What do you think?
Rate this book
How to Lie with Statistics
Darrell Huff runs the gamut of every popularly used type of statistic, probes such things as the sample study, the tabulation method, the interview technique, or the way the results are derived from the figures, and points up the countless number of dodges which are used to fool rather than to inform.
142 pages, Paperback
First published January 1, 1954
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 1,655 reviews
September 11, 2011
This is one of those rare books I would recommend to almost anyone. It's clear, concise, funny, not too complex, and above all important for anyone who wants to understand politics, economics, science, or life in general. Statistical analysis is so vital to determining how things actually stand and where we should be moving that people lacking awareness of basic logical/statistical fallacies are doomed to live within delusions. Being informed necessitates understanding and being skeptical of statistics.
Yes, the book is a bit dated, but it didn't bother me in the least. So you have to increase the monetary figures to adjust for inflation, big deal. The lessons in this book are timeless.
Also, if you do want to learn how to overstate your case or misrepresent something, this book can help. It's really written to arm people with the right questions and a healthy dose of skepticism, but inevitably ends up helping the scammers as well. Making an average? Choose the one (mean, mode, median, etc.) that best represents your case. Showing a trend over time? Choose a base year that skews things the way you want. It's not hard.
Yes, the book is a bit dated, but it didn't bother me in the least. So you have to increase the monetary figures to adjust for inflation, big deal. The lessons in this book are timeless.
Also, if you do want to learn how to overstate your case or misrepresent something, this book can help. It's really written to arm people with the right questions and a healthy dose of skepticism, but inevitably ends up helping the scammers as well. Making an average? Choose the one (mean, mode, median, etc.) that best represents your case. Showing a trend over time? Choose a base year that skews things the way you want. It's not hard.
April 30, 2014
Lies, Damn Lies, and Statistics: The Pirates of the Powerpoint
Darrell Huff uses a simple, but effective literary device to impress his readers about how much statistics affect their daily lives and their understanding of the world.
He does this by pretending that the book is a sort of primer in ways to use statistics to deceive, like a manual for swindlers, or better, for pirates. He then pretends to justify the crookedness of the book in the manner of the retired burglar whose published reminiscences amounted to a graduate course in how to pick a lock and muffle a footfall: The crooks already know these tricks; honest men must learn them in self-defense.
This keeps the book interesting and entertaining, though for anyone even partly trained in statistics, it has very little educational value.
Of course, the title of this book and Huff’s little charade would seem to imply that all such operations are the product of intent to deceive. The intelligent reader would be skeptical — it is the unfortunate truth that it not chicanery much of the time, but incompetence. On the other hand, Huff is pretty clear that the ‘errors’ if that is what they are always seem to come down on the side of the interested party. As long as the errors remain one-sided, he says, it is not easy to attribute them to bungling or accident.
No More Mr. Nice Guy
After being fellow pirates for much of the book, in the concluding chapter Huff finally lets go if his pet charade and faces up to the more serious purpose of the book: explaining how to look a phony statistic in the eye and face it down; and no less important, how to recognize sound and usable data in that wilderness of fraud to which the previous chapters have been largely devoted. He lays down some thumb rules, which in the end comes come down to asking intelligent questions of the stats, especially of the conclusions. Asking such questions require the readers to be aware of the tendency of stats to mislead and to not be dazzled by the numbers.
Huff’s book is primarily an attempt to pull down the high estimation automatically awarded to anybody willing to quote numbers. It is a fun evening read for the expert, who may then roll his eyes and say that there is nothing of real value in the book. But as its popularity attests to, it seems to be an important book for the lay reader, just by serving a reminder that the pirates are still out there — wielding their charts.
August 3, 2021
3.5
ترجمه‌ای که توسط "نشر دنیای اقتصاد" از این کتاب ارائه شده رو نخونید. مشکلاتی بنظر من داره که توی پانویس نوشتم.
به طور کلی، کتاب در تلاشه به شما نشون بده که آمار، علاوه بر علم، یک جور هنر(!) هم محسوب می‌شه و از بدترین و اسفناک‌ترین شرایط می‌شه آمار درخشانی بیرون کشید یا بالعکس، یک شرایط عادی و حتی خوب رو فاجعه‌بار یا بحرانی نشون داد. من اینجا سعی می‌کنم خلاصه‌ای از کتاب ارائه کنم تا اگر کسی حوصله نداشت یا وقت نداشت بتونه از همین استفاده کنه. خصوصا برای خودم مهمه که چیزکی نوشته باشم تا همگی بتونیم این چند نکته ساده اما کلیدی رو با هم مرور کنیم تا خوب در خاطرمون حک بشه:
1-
آماری رو در نظر بگیرید که می‌گه، متوسط جمعیت فلان منطقه، 3.6 نفر به ازای هر خانوار هستش. حالا یک شرکت انبوه‌ساز، بدون مطالعه و بر مبنای این آمار میاد و تعداد زیادی خونه 2 خوابه توی اون منطقه می‌سازه و بعد از مدتی، به خاک سیاه می‌شینه! چرا؟
با مطالعه بیشتر جمعیت منطقه متوجه می‌شیم که اگرچه متوسط جمعیت اون منطقه 3.6 نفره، اما 45% خانواده‌های منطقه، سه یا چهار نفری، 35% یک یا دو نفری و 20% پنج نفر به بالا هستن. لذا برای بیش از نیمی از خانواده‌های این منطقه، خانه‌های ساخته شده مناسب نیست! (اینطوری حساب کنید که اثرات خانواده‌های 1 و 2 نفری با اثر خانواده‌های 5 نفر به بالا به نوعی همدیگه رو خنثی کردن)
چیزی که در آمار اولیه مشهود نیست، نحوه پراکندگی داده‌ها هستن. این یکی از متداول‌ترین مواردی که هستش که عمدتا به طور عمدی و بعضا به طور سهوی باعث خطای مخاطبین آمار می‌شه. در اخبار یا تبلیغات بازرگانی این مساله بسیار متداوله که آماری به شما ارائه بشه بدون اینکه توضیحی از کم و کیف داده‌های خام داده بشه. به این شکل از آمار هیچ وقت اعتماد نکنید.
مثلا ممکنه یک تیتر خبری ببینیم یا بشنویم راجع به رشد صادرات ارزی یا رشد مثبت اقتصادی. بدون اینکه بدونیم این رشد ناشی از چیه و کدوم بخش‌ها صعودی بوده و کدوم بخش‌ها (احیانا صنعت) نزولی بوده.
2-
گاهی اما پنهانکاری و دروغ، ناشی از گویا نبودن معیار ارائه شده (مثلا صرف گفتن مقدار میانگین) نیست. بلکه در مورد تناسب اندازه نمونه است. این ترفند بیشتر در تبلیغات بازرگانی متداوله. مثلا در تبلیغ تلویزیونی مربوط به یک خمیر دندان ادعا می‌شه که با خاصیت جلوگیری از پوسیدگی دندان به میزان 60 درصد!! (بعد هم یک پیکان قرمز گنده رو نشون میده که از 100 میاد روی 40 و حسابی شما رو تحت تاثیر قرار میده)
اما داستان اینه که کلا 20 نفر رو بررسی کردن و 12 نفر اونها بعد از مدت زمان انجام آزمایش (که باز خود این مدت زمان محل مناقشه است) پوسیدگی دندونشون بیشتر نشده. به این ترتیب نتیجه‌گیری شده که استفاده از این محصول تا 60 درصد در جلوگیری از پوسیدگی موثره. مساله‌ای که اینجا وجود داره اینه که برای چنین مطالعاتی، اندازه نمونه باید بسیار بزرگتر باشه. اما در اون صورت احتمالا آمار چندان جالب توجهی برای امور تبلیغاتی حاصل نمی‌شد.
اصولا همیشه این مثال رو به یاد داشته باشید: یک سکه رو 100 بار بالا می‌اندازیم. 44 بار شیر میاد و56 بار خط. آیا این آزمایش، که توسط عالی‌ترین سازمان‌های نظارتی به تایید رسیده، هیچ معنایی داره؟ نه. این مساله هرچند بدیهی به نظر میاد، اما به کرات و حتی در مطالعات علمی هم بسیار رایجه. پس در موقع خوندن و بررسی هر آمار و ارقامی به یاد داشته باشید که اندازه نمونه با نوع مطالعه تناسب داشته باشه و اگر در تشخیص اون متخصص نیستید، همیشه به دیده شک نگاه کنید.
3-
مورد سومی که قابل توجهه، اینه که تشخیص بدیم آیا آمار و ارقام ارائه شده رابطه علّی با نتیجه‌گیری داره یا نه. این هم یکی از اون مواردی هستش که بسیار خطرناکه و خیلی پر کاربرده.
مثلا باز در یک آگهی بازرگانی ادعا می‌شه که آیا می‌دانید 43 درصد از دندان‌پزشکان از مسواک فلان استفاده میکنن؟! خیلی تاثیرگذاره. نه؟ اما واقعیت اینه که هیچ ارتباط خاصی بین استفاده بیشتر دندان‌پزشکان از یک مسواک بخصوص و موثرتر بودن اون وجود نداره و اساسا نوع مسواک اگر نه هیچ، که تفاوت بسیار قلیلی از لحاظ نگه‌داری دندان‌ها رقم میزنن! قاعدتا اونچه مهمه درست و مستمر مسواک زدنه، نه برند مسواک. چه بسا بسیاری از دندان‌پزشکان به خاطر درآمد بالا و سبک زندگی که دارن، از اون برند گرون قیمت استفاده میکنن.
مثال دیگه از نوع غیر تبلیغاتی: طی یک مطالعه آماری که توسط پژوهشکده فلان روی 1500 نفر از فارغ التحصیلان دانشگاه‌های تهران با سن حدود 40 سال انجام شد، مشخص شده که 89 درصد از فارغ‌التحصیلان مرد موفق به ازدواج شدن (حال آنکه این درصد برای کل جمعیت مردان 40 ساله تهران، 80 درصده)، و از میان فارغ‌التحصیلان زن، 78 درصد موفق به ازدواج شدن (در حالی که آمار کلی برای زنان تهرانی 94 درصده). بر مبنای این مطالعه، چنین نتیجه‌گیری شده که رفتن به دانشگاه شانس ازدواج مردان رو افزایش و شانس ازدواج خانم‌ها رو کاهش میده
{در مثال مناقشه نکنید :)) ا}
مشکل این مثال هم باز شبیه مثال قبلیه اگرچه کمتر مشهوده. هرچند آمار قابل اعتماد و معتبره، اما این اعداد صرفا یک همبستگی* رو نشون می‌دن و ابدا هیچ معنایی از جنس علت-معلولی نمی‌شه از اونها استخراج کرد (برای این کار آمار بسیار بیشتر و شرایط کنترل شده بیشتری نیازه). چه بسا این دسته از زنانی که به دانشگاه رفتن، حتی اگر هم تحصیل نمی‌کردن، از شانس ازدواج پایین‌تری برخوردار بودن. و به همین ترتیب، برای مردان.
* Correlation
این دقیقا مثل اینه که آمار سیگار کشیدن بین دانش‌آموزان رو در بیاریم و بخوایم دلیل درس نخوندن اونها رو به سیگار کشیدن نسبت بدیم. هرچند همبستگی قوی‌ای مشاهده می‌شه در این موارد، اما غالبا اینها هر دو معلول یک عامل سوم (و مطالعه نشده) دیگه‌ای هستن که معمولا مشکلات خانوادگی، اعتیاد والدین و مواردی از این قبیل هستش. لذا همیشه حواستون باشه همبستگی رو با رابطه علت-معلولی اشتباه نگیرید.
4-
آخرین نکته، دقت مطالعه است. همیشه نسبت به دقت اعدادی که داره بهتون ارائه می‌شه هشیار و مشکوک باشید. بهترین و صادقانه‌ترین مطالعات و آمارها، باید میزان خطای احتمالی رو ارائه کنن. برای مثال، وقتی دقیق‌ترین محققین علوم رفتاری می‌خوان سهم اثرات ژنتیکی در رفتار فرد رو تقریب بزنن، میگن بین 30 تا 60 درصد (منظورم اینه که یک کار که بخواد حسابی علمی بررسی بشه، چقدر باید جای خطا توش لحاظ کرد)
حالا فرض کنید چه اثری می‌گذاشت روی شما اگر گفته می‌شد، بین 32.7 تا 61.3 درصد؟ بازه همون بازه است. اما این بار این حس به آدم القا می‌شه که با مطالعه دقیقتری روبرو هستیم. در حالی که چنین دقتی اینجا اصلا بی‌معنیه و اگر اصولی بخوایم برخورد کنیم، اساسا چنین اعدادی حق ندارن بکار برده بشن، چون خارج از دقت اندازه‌گیری ماست و فاقد مفهومه. (مثل این می‌مونه که مثلا شما با خط کشی که دقتش 1 میلی‌متر هست، سه عدد رو اندازه بگیری و متوسطش رو 10.75 میلی‌متر عنوان کنی. این اعشار با توجه به دقت ابزار شما اساسا بی‌معنیه). با اینحال، این دقت همچنان ما رو می‌فریبه و ابزار خوبی برای دروغگویی و القای دقت و صحت کار محسوب می‌شه.
مثلا گفته می‌شه میانگین خواب در تهران برای مردان 7 ساعت و 37 دقیقه و برای زنان 8 ساعت و 11 دقیقه است. حتی ممکنه پا رو از این هم فراتر بذارن و بر مبنای این مطالعه ادعا کنن که بدن زنان بیشتر از مردان نیاز به خواب داره. (از این دست مقایسه‌ها خصوصا در مقایسه���های رفتاری بین زنان و مردان تا دلتون بخواد هست)
مساله‌ای که هست اینه که شما وقتی در مورد ساعت خواب کسی سوال کنی، بهترین و دقیق‌ترین افراد هم به سختی بتونن دقتی فراتر از نیم ساعت- نه، یک ربع- به شما بدن. تازه این مشروط به اینه که شما در برگه پرسشنامه، جای خالی برای ساعت خواب گذاشته باشی و چند گزینه‌ای نباشه. لذا این دقت برای این اعداد اساسا بی‌معنیه. و نمی‌تونه و نباید مبنای مقایسه قرار بگیره. خصوصا زمانی که اختلافها مثل این مثال، اونقدر قابل توجه نیست و کل نتیجه‌گیری می‌تونه کاملا ناشی از خطای ارائه نتایج باشه.
کتاب البته نکات زیاد دیگه‌ای رو هم گفته. من جمله کلک‌هایی که توی نمودارها میزنن تا روند تغییرات رو اونطور که مطلوبه نشون بدن. اما بگمانم خواننده ایرانی اونقدر دروغ دیده و شنیده که اونا دیگه خیلی بچه‌بازی هستش برا ما! اما سر جمع بنظرم کتاب می‌تونست خیلی بهتر باشه و حالا که بحث رو اینقدر خوب باز کرده، کاش حرفه‌ای‌تر و همه جانبه‌تر به بحث پرداخته بود. اما خوب، هدف کتاب طبقه عام جامعه بوده و با توجه به اینکه کتاب خیلی قدیمیه، نمی‌شه انتظار زیادی هم داشت.
پانویس
اندکی راجع به ترجمه
متاسفانه اونچه که از اون به نام تالیف و ترجمه یاد شده، تفاوتهایی ورای اضافه کردن چند مثال امروزی به متن اصلی کتابه و متاسفانه مترجم تا حد زیادی گرایشهای سیاسی خودش رو وارد کتاب کرده و کتاب رو از اعتبار ساقط کرده.
سرفصل ها بی هیچ دلیلی جابجا شدن و برخی توضیحات حذف شدن و در عوض برخی فصلهای جدید (من جمله فصل 3) به کتاب اضافه شده که طی اون متاسفانه مترجم-مولف محترم با رویکردی کاملا ژورنالیستی و بدون اشراف کامل بر منطق فازی اومده و مطالبی رو عنوان کرده که فقط ظاهر علمی داره و طی اون ایشون اومدن همه شاخص های جهانی رو زیر سوال بردن و استدلال میکنن که مثلا فساد یک مفهوم فازی هستش لذا اینکه طبق شاخصهای جهانی کشور الف در رتبه بندی فساد رتبه 100 داره و در همسایگی اون کشور ب رتبه 20 هست، از نظر مولف بی معنیه چون فرهنگ اون دو تا کشور بسیار به هم نزدیکه و این رتبه بندی ها اولا ممکنه مغرضانه باشه چون کشورهای زیادی در دنیا با کشور الف مشکل دارن، و دوم اینکه بخشی از فساد ممکنه وجود داشته باشه از نگاه ناظرین دور مونده باشه.
دیگه ببینید خودتون خلط مبحث رو و تلفیق واژگان علمی و کلام غیر علمی رو. استدلالهای درست و استدلالهای غلط رو. مترجم-مولف ما که خودش از دور دستی بر سیاست داره، سعی داره مسائل مهمی رو کم اهمیت جلوه بده و برخلاف کتاب زبان اصلی که سعی در شفافسازی و طبقه بندی ذهن شما داره، تلاش این کتاب ترجمه-تالیفی ایجاد بی اعتمادی به همه آمار ها و ارقامه تا از این طریق شاید بتونه رتبه های نامناسب کشورمون در شاخصهای جهانی نظیر فلاکت، خوشحالی، فساد و... رو توجیه کنه. حقیقتا من دلیلی برای جابجایی سرفصلها و ادغام و هم زدن اونها با هم پیدا نکردم. متن زبان اصلی خیلی خوب طبقه بندی شده بود. جالب توجه اینکه مولف ما برای این منظور اول میاد خودش رو در قالب منتقد سیاستهای دولتهای پیشین قرار میده تا بیشتر اعتماد خواننده رو جلب کنه. بیش از این حرفی نیست
ترجمه‌ای که توسط "نشر دنیای اقتصاد" از این کتاب ارائه شده رو نخونید. مشکلاتی بنظر من داره که توی پانویس نوشتم.
به طور کلی، کتاب در تلاشه به شما نشون بده که آمار، علاوه بر علم، یک جور هنر(!) هم محسوب می‌شه و از بدترین و اسفناک‌ترین شرایط می‌شه آمار درخشانی بیرون کشید یا بالعکس، یک شرایط عادی و حتی خوب رو فاجعه‌بار یا بحرانی نشون داد. من اینجا سعی می‌کنم خلاصه‌ای از کتاب ارائه کنم تا اگر کسی حوصله نداشت یا وقت نداشت بتونه از همین استفاده کنه. خصوصا برای خودم مهمه که چیزکی نوشته باشم تا همگی بتونیم این چند نکته ساده اما کلیدی رو با هم مرور کنیم تا خوب در خاطرمون حک بشه:
1-
آماری رو در نظر بگیرید که می‌گه، متوسط جمعیت فلان منطقه، 3.6 نفر به ازای هر خانوار هستش. حالا یک شرکت انبوه‌ساز، بدون مطالعه و بر مبنای این آمار میاد و تعداد زیادی خونه 2 خوابه توی اون منطقه می‌سازه و بعد از مدتی، به خاک سیاه می‌شینه! چرا؟
با مطالعه بیشتر جمعیت منطقه متوجه می‌شیم که اگرچه متوسط جمعیت اون منطقه 3.6 نفره، اما 45% خانواده‌های منطقه، سه یا چهار نفری، 35% یک یا دو نفری و 20% پنج نفر به بالا هستن. لذا برای بیش از نیمی از خانواده‌های این منطقه، خانه‌های ساخته شده مناسب نیست! (اینطوری حساب کنید که اثرات خانواده‌های 1 و 2 نفری با اثر خانواده‌های 5 نفر به بالا به نوعی همدیگه رو خنثی کردن)
چیزی که در آمار اولیه مشهود نیست، نحوه پراکندگی داده‌ها هستن. این یکی از متداول‌ترین مواردی که هستش که عمدتا به طور عمدی و بعضا به طور سهوی باعث خطای مخاطبین آمار می‌شه. در اخبار یا تبلیغات بازرگانی این مساله بسیار متداوله که آماری به شما ارائه بشه بدون اینکه توضیحی از کم و کیف داده‌های خام داده بشه. به این شکل از آمار هیچ وقت اعتماد نکنید.
مثلا ممکنه یک تیتر خبری ببینیم یا بشنویم راجع به رشد صادرات ارزی یا رشد مثبت اقتصادی. بدون اینکه بدونیم این رشد ناشی از چیه و کدوم بخش‌ها صعودی بوده و کدوم بخش‌ها (احیانا صنعت) نزولی بوده.
2-
گاهی اما پنهانکاری و دروغ، ناشی از گویا نبودن معیار ارائه شده (مثلا صرف گفتن مقدار میانگین) نیست. بلکه در مورد تناسب اندازه نمونه است. این ترفند بیشتر در تبلیغات بازرگانی متداوله. مثلا در تبلیغ تلویزیونی مربوط به یک خمیر دندان ادعا می‌شه که با خاصیت جلوگیری از پوسیدگی دندان به میزان 60 درصد!! (بعد هم یک پیکان قرمز گنده رو نشون میده که از 100 میاد روی 40 و حسابی شما رو تحت تاثیر قرار میده)
اما داستان اینه که کلا 20 نفر رو بررسی کردن و 12 نفر اونها بعد از مدت زمان انجام آزمایش (که باز خود این مدت زمان محل مناقشه است) پوسیدگی دندونشون بیشتر نشده. به این ترتیب نتیجه‌گیری شده که استفاده از این محصول تا 60 درصد در جلوگیری از پوسیدگی موثره. مساله‌ای که اینجا وجود داره اینه که برای چنین مطالعاتی، اندازه نمونه باید بسیار بزرگتر باشه. اما در اون صورت احتمالا آمار چندان جالب توجهی برای امور تبلیغاتی حاصل نمی‌شد.
اصولا همیشه این مثال رو به یاد داشته باشید: یک سکه رو 100 بار بالا می‌اندازیم. 44 بار شیر میاد و56 بار خط. آیا این آزمایش، که توسط عالی‌ترین سازمان‌های نظارتی به تایید رسیده، هیچ معنایی داره؟ نه. این مساله هرچند بدیهی به نظر میاد، اما به کرات و حتی در مطالعات علمی هم بسیار رایجه. پس در موقع خوندن و بررسی هر آمار و ارقامی به یاد داشته باشید که اندازه نمونه با نوع مطالعه تناسب داشته باشه و اگر در تشخیص اون متخصص نیستید، همیشه به دیده شک نگاه کنید.
3-
مورد سومی که قابل توجهه، اینه که تشخیص بدیم آیا آمار و ارقام ارائه شده رابطه علّی با نتیجه‌گیری داره یا نه. این هم یکی از اون مواردی هستش که بسیار خطرناکه و خیلی پر کاربرده.
مثلا باز در یک آگهی بازرگانی ادعا می‌شه که آیا می‌دانید 43 درصد از دندان‌پزشکان از مسواک فلان استفاده میکنن؟! خیلی تاثیرگذاره. نه؟ اما واقعیت اینه که هیچ ارتباط خاصی بین استفاده بیشتر دندان‌پزشکان از یک مسواک بخصوص و موثرتر بودن اون وجود نداره و اساسا نوع مسواک اگر نه هیچ، که تفاوت بسیار قلیلی از لحاظ نگه‌داری دندان‌ها رقم میزنن! قاعدتا اونچه مهمه درست و مستمر مسواک زدنه، نه برند مسواک. چه بسا بسیاری از دندان‌پزشکان به خاطر درآمد بالا و سبک زندگی که دارن، از اون برند گرون قیمت استفاده میکنن.
مثال دیگه از نوع غیر تبلیغاتی: طی یک مطالعه آماری که توسط پژوهشکده فلان روی 1500 نفر از فارغ التحصیلان دانشگاه‌های تهران با سن حدود 40 سال انجام شد، مشخص شده که 89 درصد از فارغ‌التحصیلان مرد موفق به ازدواج شدن (حال آنکه این درصد برای کل جمعیت مردان 40 ساله تهران، 80 درصده)، و از میان فارغ‌التحصیلان زن، 78 درصد موفق به ازدواج شدن (در حالی که آمار کلی برای زنان تهرانی 94 درصده). بر مبنای این مطالعه، چنین نتیجه‌گیری شده که رفتن به دانشگاه شانس ازدواج مردان رو افزایش و شانس ازدواج خانم‌ها رو کاهش میده
{در مثال مناقشه نکنید :)) ا}
مشکل این مثال هم باز شبیه مثال قبلیه اگرچه کمتر مشهوده. هرچند آمار قابل اعتماد و معتبره، اما این اعداد صرفا یک همبستگی* رو نشون می‌دن و ابدا هیچ معنایی از جنس علت-معلولی نمی‌شه از اونها استخراج کرد (برای این کار آمار بسیار بیشتر و شرایط کنترل شده بیشتری نیازه). چه بسا این دسته از زنانی که به دانشگاه رفتن، حتی اگر هم تحصیل نمی‌کردن، از شانس ازدواج پایین‌تری برخوردار بودن. و به همین ترتیب، برای مردان.
* Correlation
این دقیقا مثل اینه که آمار سیگار کشیدن بین دانش‌آموزان رو در بیاریم و بخوایم دلیل درس نخوندن اونها رو به سیگار کشیدن نسبت بدیم. هرچند همبستگی قوی‌ای مشاهده می‌شه در این موارد، اما غالبا اینها هر دو معلول یک عامل سوم (و مطالعه نشده) دیگه‌ای هستن که معمولا مشکلات خانوادگی، اعتیاد والدین و مواردی از این قبیل هستش. لذا همیشه حواستون باشه همبستگی رو با رابطه علت-معلولی اشتباه نگیرید.
4-
آخرین نکته، دقت مطالعه است. همیشه نسبت به دقت اعدادی که داره بهتون ارائه می‌شه هشیار و مشکوک باشید. بهترین و صادقانه‌ترین مطالعات و آمارها، باید میزان خطای احتمالی رو ارائه کنن. برای مثال، وقتی دقیق‌ترین محققین علوم رفتاری می‌خوان سهم اثرات ژنتیکی در رفتار فرد رو تقریب بزنن، میگن بین 30 تا 60 درصد (منظورم اینه که یک کار که بخواد حسابی علمی بررسی بشه، چقدر باید جای خطا توش لحاظ کرد)
حالا فرض کنید چه اثری می‌گذاشت روی شما اگر گفته می‌شد، بین 32.7 تا 61.3 درصد؟ بازه همون بازه است. اما این بار این حس به آدم القا می‌شه که با مطالعه دقیقتری روبرو هستیم. در حالی که چنین دقتی اینجا اصلا بی‌معنیه و اگر اصولی بخوایم برخورد کنیم، اساسا چنین اعدادی حق ندارن بکار برده بشن، چون خارج از دقت اندازه‌گیری ماست و فاقد مفهومه. (مثل این می‌مونه که مثلا شما با خط کشی که دقتش 1 میلی‌متر هست، سه عدد رو اندازه بگیری و متوسطش رو 10.75 میلی‌متر عنوان کنی. این اعشار با توجه به دقت ابزار شما اساسا بی‌معنیه). با اینحال، این دقت همچنان ما رو می‌فریبه و ابزار خوبی برای دروغگویی و القای دقت و صحت کار محسوب می‌شه.
مثلا گفته می‌شه میانگین خواب در تهران برای مردان 7 ساعت و 37 دقیقه و برای زنان 8 ساعت و 11 دقیقه است. حتی ممکنه پا رو از این هم فراتر بذارن و بر مبنای این مطالعه ادعا کنن که بدن زنان بیشتر از مردان نیاز به خواب داره. (از این دست مقایسه‌ها خصوصا در مقایسه���های رفتاری بین زنان و مردان تا دلتون بخواد هست)
مساله‌ای که هست اینه که شما وقتی در مورد ساعت خواب کسی سوال کنی، بهترین و دقیق‌ترین افراد هم به سختی بتونن دقتی فراتر از نیم ساعت- نه، یک ربع- به شما بدن. تازه این مشروط به اینه که شما در برگه پرسشنامه، جای خالی برای ساعت خواب گذاشته باشی و چند گزینه‌ای نباشه. لذا این دقت برای این اعداد اساسا بی‌معنیه. و نمی‌تونه و نباید مبنای مقایسه قرار بگیره. خصوصا زمانی که اختلافها مثل این مثال، اونقدر قابل توجه نیست و کل نتیجه‌گیری می‌تونه کاملا ناشی از خطای ارائه نتایج باشه.
کتاب البته نکات زیاد دیگه‌ای رو هم گفته. من جمله کلک‌هایی که توی نمودارها میزنن تا روند تغییرات رو اونطور که مطلوبه نشون بدن. اما بگمانم خواننده ایرانی اونقدر دروغ دیده و شنیده که اونا دیگه خیلی بچه‌بازی هستش برا ما! اما سر جمع بنظرم کتاب می‌تونست خیلی بهتر باشه و حالا که بحث رو اینقدر خوب باز کرده، کاش حرفه‌ای‌تر و همه جانبه‌تر به بحث پرداخته بود. اما خوب، هدف کتاب طبقه عام جامعه بوده و با توجه به اینکه کتاب خیلی قدیمیه، نمی‌شه انتظار زیادی هم داشت.
پانویس
اندکی راجع به ترجمه
متاسفانه اونچه که از اون به نام تالیف و ترجمه یاد شده، تفاوتهایی ورای اضافه کردن چند مثال امروزی به متن اصلی کتابه و متاسفانه مترجم تا حد زیادی گرایشهای سیاسی خودش رو وارد کتاب کرده و کتاب رو از اعتبار ساقط کرده.
سرفصل ها بی هیچ دلیلی جابجا شدن و برخی توضیحات حذف شدن و در عوض برخی فصلهای جدید (من جمله فصل 3) به کتاب اضافه شده که طی اون متاسفانه مترجم-مولف محترم با رویکردی کاملا ژورنالیستی و بدون اشراف کامل بر منطق فازی اومده و مطالبی رو عنوان کرده که فقط ظاهر علمی داره و طی اون ایشون اومدن همه شاخص های جهانی رو زیر سوال بردن و استدلال میکنن که مثلا فساد یک مفهوم فازی هستش لذا اینکه طبق شاخصهای جهانی کشور الف در رتبه بندی فساد رتبه 100 داره و در همسایگی اون کشور ب رتبه 20 هست، از نظر مولف بی معنیه چون فرهنگ اون دو تا کشور بسیار به هم نزدیکه و این رتبه بندی ها اولا ممکنه مغرضانه باشه چون کشورهای زیادی در دنیا با کشور الف مشکل دارن، و دوم اینکه بخشی از فساد ممکنه وجود داشته باشه از نگاه ناظرین دور مونده باشه.
دیگه ببینید خودتون خلط مبحث رو و تلفیق واژگان علمی و کلام غیر علمی رو. استدلالهای درست و استدلالهای غلط رو. مترجم-مولف ما که خودش از دور دستی بر سیاست داره، سعی داره مسائل مهمی رو کم اهمیت جلوه بده و برخلاف کتاب زبان اصلی که سعی در شفافسازی و طبقه بندی ذهن شما داره، تلاش این کتاب ترجمه-تالیفی ایجاد بی اعتمادی به همه آمار ها و ارقامه تا از این طریق شاید بتونه رتبه های نامناسب کشورمون در شاخصهای جهانی نظیر فلاکت، خوشحالی، فساد و... رو توجیه کنه. حقیقتا من دلیلی برای جابجایی سرفصلها و ادغام و هم زدن اونها با هم پیدا نکردم. متن زبان اصلی خیلی خوب طبقه بندی شده بود. جالب توجه اینکه مولف ما برای این منظور اول میاد خودش رو در قالب منتقد سیاستهای دولتهای پیشین قرار میده تا بیشتر اعتماد خواننده رو جلب کنه. بیش از این حرفی نیست
November 10, 2019
This little book was first published in the Fifties and has remained in print even as the cover cost and the examples of merchandise in the book have been updated for inflation. Why? Because the principles it teaches are just as important now as then. See how government, big business, pressure groups and labor all manipulate us with number-mangling to indicate changes in prices, wages or unemployment are better or worse than they really are, or how the government's policy is the right one even though it may be terrible (and on the other hand, how the opponent uses statistics to make things gloomier than reality).
Written with a smile, easily understandable, yet with a fine sense of how statistics can be used against us even, allegedly, "for our own good." Every American should have a copy of HOW TO LIE WITH STATISTICS. After this book, if you'd like to get further into the nitty-gritty of numerical manipulation, how ordinary Americans are routinely deceived by it, and what you can do about it, consider INNUMERACY by John Allen Paulos.
Written with a smile, easily understandable, yet with a fine sense of how statistics can be used against us even, allegedly, "for our own good." Every American should have a copy of HOW TO LIE WITH STATISTICS. After this book, if you'd like to get further into the nitty-gritty of numerical manipulation, how ordinary Americans are routinely deceived by it, and what you can do about it, consider INNUMERACY by John Allen Paulos.
January 15, 2023
2021-03-27 Remembering this book yet again, since it is truly "One of the Best books I have ever read." If you understand this book, and it ain't hard, and you apply it to the statistics you come across in your life, then you have a big leg up on NOT being bamboozled.
The book is really short, big print, lots of illustrations, great humor and absolutely a CLASSIC.
If you had a tough time in the statistics section of your High School math class, read this to get the basics - you need it and you will benefit from it.
This was a supplementary text to my college statistics class and wow, was I glad it was. That course was not easy. Logical to the max, and useful, but it went into more depth and detail and agony of factoring and other difficult/tedious work than I can say really helped me. But this book! The simple lessons in it have stayed with me, and animate any presentation, book, technical article, advertisement, political speech, etc. etc. that I come across that uses statistics.
What could be better than that?
Oh, and did I mention that I have bought, given away and recommended this book more than almost any other book I have read?
The book is really short, big print, lots of illustrations, great humor and absolutely a CLASSIC.
If you had a tough time in the statistics section of your High School math class, read this to get the basics - you need it and you will benefit from it.
This was a supplementary text to my college statistics class and wow, was I glad it was. That course was not easy. Logical to the max, and useful, but it went into more depth and detail and agony of factoring and other difficult/tedious work than I can say really helped me. But this book! The simple lessons in it have stayed with me, and animate any presentation, book, technical article, advertisement, political speech, etc. etc. that I come across that uses statistics.
What could be better than that?
Oh, and did I mention that I have bought, given away and recommended this book more than almost any other book I have read?
September 19, 2023
The best-selling (popular) Statistics book…
--This book aims to make stats accessible for the public. Thus, it targets:
a) Those who get lost in/avoid statistics for fear of mathematics and/or ignorance of its application in social issues.
…and converts them to:
b) Those who bother with statistics not for its mathematics, but for its application in social issues.
--Going easy on the math is not actually sacrificing too much. Consider: public deception requires the victim thinking they understand something, a bait-and-switch. Thus, complex math would often be counter-productive!
--The deception is in the lure of “simple” math:
-ex. “1 in 3 people…!”
-ex. “An increase of 200%!”
-ex. “Look at this graph with a dramatic, steep curve!”
...you must be first convinced that you understand it, while the logic behind it is manipulated...
--As for the folks who love abstract technicalities and shy away from the noisy world of human irrationality and power struggles, you’ll have no problem learning the mathematics of statistics elsewhere.
...For the rest of us, this book is the ideal starting point. The excellent Ben Goldacre takes this to another level by applying this approach to science/public health/media/policy: "Pulling bad science apart is the best teaching gimmick I know for explaining how good science really works."
-Bad Science: Quacks, Hacks, and Big Pharma Flacks
-I Think You'll Find It's a Bit More Complicated Than That
-Bad Pharma: How Drug Companies Mislead Doctors and Harm Patients
Highlights:
1) Representative sample?
--We are sold conclusions with authoritative precision (“33.33% of users…!”), but first consider how the sample (subset of the total population used as an estimate, since surveying total population is usually impossible) was determined:
a) Often the sample is simply too small (sampled 3 people, found 1!), so the results are biased by chance. As always, if the raw numbers of the sample size are not even provided, be suspicious! If they are given (in fine print), often a cursory comparison of the sample size vs. actual population size is enough to spot the lie.
b) More mathematical checks are mentioned only in a cursory manner and need to be supplemented elsewhere. These include the statistical significance level for handling false positives (which has limitations: effect size and reproducibility), reliability of the correlation (standard deviation/standard error…), etc.
c) Next, many methodological layers (ex. with data collection) need to be unraveled to detect further sampling biases and non-sampling errors. And as always, the first suspicious sign is if the methodological details are hidden.
2) Which “average”?
--Different types have different perspectives/results; this can be unspecified or used inconsistently in comparisons: “mean” hides outliers that skew the result (think mean income where one CEO's salary skews the many low-wage workers); “median” (middle value) and “mode” (most common value) each provide insights and limitations for manipulation (or just sloppy misinterpretation).
3) What are the visuals presenting and hiding?
--Just as simple math is a lure, simple visuals are full of tricks. Hidden data, ranges cut off, distorted proportions of the units of the x/y axis, distorted perspectives (ex. coloured voting map which emphasizes land mass instead of population density), etc.
4) What is the logic behind the interpretation?
a) A key methodological trick is the bait-and-switch, by making overreaching claims based on a “semi-attached figure”. This is very common in health/medical claims (see Goldacre's books above), where a surrogate outcome (ex. a singular result from an isolated chemical reaction in a lab Petri dish) is falsely extrapolated as having an equal impact on a real clinical outcome (i.e. on a complex condition in a complex human body in a complex social environment).
...This builds up to tremendous distortions since private companies can simply not publish negative findings (the enforcement to pre-register clinical trials is abysmal), thus creating systemic publication bias!
b) Quick glance at logical fallacies (esp. regarding correlation vs. causation). A popular one is post hoc ergo propter hoc (if B follows A, then assume A caused B).
--Ex. Many people who get lung cancer drink alcohol, so drinking alcohol must cause lung cancer...? (Actually, many people who drink alcohol also smoke...)
c) The illusion of the shifting base: Ex. advertising a whooping 100% savings on $100... so it's free? Well, it turns out the savings was actually 50%. The "100% savings" is comparing the new (reduced) price of $50 to the savings ($50), when it should be comparing with the original base price of $100.
...This can get really sneaky the other way around, ex. locked into paying interest routinely on the original loan amount whereas you should pay on the new decreasing amount (since you've been paying off your loan)!
d) Plenty of counter-intuitive tricks can be done with percentages. Goldacre recommends using real numbers (1 out of 300) instead of percentages if you want to communicate more clearly/intuitively (esp. to describe changes). After all, an 100% increase could just be a change from 1-in-1,000,000 to 2-in-1,000,000.
The Missing:
--I have to add this note, because I’ve seen Bill Gates recommend this book (and novice conspiracy hacks getting the wrong ideas). Despite this book providing technical tools to spot manipulation, it presents “power” as an abstract bogeyman (mostly in the form of petty con artists).
1) Systemic power:
--Unlike the misleading narratives of petty con artists being “bad apples”, systemic power thrives in abstraction; this is how it builds social consent.
--To understand power, we must turn to political economy. Goldacre provides a useful bridge, by connecting these statistical deceptions to power in a step-by-step approach: starting with petty con artist quacks, then expanding to bigger systems: capitalist media and industries (including Big Pharma, which avoiding sensationalism which Big Media thrives on):
-Talking to My Daughter About the Economy: or, How Capitalism Works—and How It Fails
-Another Now: Dispatches from an Alternative Present
-Understanding Power: The Indispensable Chomsky
2) Propaganda:
--To expand our analysis of propaganda:
-Necessary Illusions: Thought Control in Democratic Societies
-Manufacturing Consent: The Political Economy of the Mass Media
-Merchants of Doubt: How a Handful of Scientists Obscured the Truth on Issues from Tobacco Smoke to Global Warming
--This book aims to make stats accessible for the public. Thus, it targets:
a) Those who get lost in/avoid statistics for fear of mathematics and/or ignorance of its application in social issues.
…and converts them to:
b) Those who bother with statistics not for its mathematics, but for its application in social issues.
--Going easy on the math is not actually sacrificing too much. Consider: public deception requires the victim thinking they understand something, a bait-and-switch. Thus, complex math would often be counter-productive!
--The deception is in the lure of “simple” math:
-ex. “1 in 3 people…!”
-ex. “An increase of 200%!”
-ex. “Look at this graph with a dramatic, steep curve!”
...you must be first convinced that you understand it, while the logic behind it is manipulated...
--As for the folks who love abstract technicalities and shy away from the noisy world of human irrationality and power struggles, you’ll have no problem learning the mathematics of statistics elsewhere.
...For the rest of us, this book is the ideal starting point. The excellent Ben Goldacre takes this to another level by applying this approach to science/public health/media/policy: "Pulling bad science apart is the best teaching gimmick I know for explaining how good science really works."
-Bad Science: Quacks, Hacks, and Big Pharma Flacks
-I Think You'll Find It's a Bit More Complicated Than That
-Bad Pharma: How Drug Companies Mislead Doctors and Harm Patients
Highlights:
1) Representative sample?
--We are sold conclusions with authoritative precision (“33.33% of users…!”), but first consider how the sample (subset of the total population used as an estimate, since surveying total population is usually impossible) was determined:
a) Often the sample is simply too small (sampled 3 people, found 1!), so the results are biased by chance. As always, if the raw numbers of the sample size are not even provided, be suspicious! If they are given (in fine print), often a cursory comparison of the sample size vs. actual population size is enough to spot the lie.
b) More mathematical checks are mentioned only in a cursory manner and need to be supplemented elsewhere. These include the statistical significance level for handling false positives (which has limitations: effect size and reproducibility), reliability of the correlation (standard deviation/standard error…), etc.
c) Next, many methodological layers (ex. with data collection) need to be unraveled to detect further sampling biases and non-sampling errors. And as always, the first suspicious sign is if the methodological details are hidden.
2) Which “average”?
--Different types have different perspectives/results; this can be unspecified or used inconsistently in comparisons: “mean” hides outliers that skew the result (think mean income where one CEO's salary skews the many low-wage workers); “median” (middle value) and “mode” (most common value) each provide insights and limitations for manipulation (or just sloppy misinterpretation).
3) What are the visuals presenting and hiding?
--Just as simple math is a lure, simple visuals are full of tricks. Hidden data, ranges cut off, distorted proportions of the units of the x/y axis, distorted perspectives (ex. coloured voting map which emphasizes land mass instead of population density), etc.
4) What is the logic behind the interpretation?
a) A key methodological trick is the bait-and-switch, by making overreaching claims based on a “semi-attached figure”. This is very common in health/medical claims (see Goldacre's books above), where a surrogate outcome (ex. a singular result from an isolated chemical reaction in a lab Petri dish) is falsely extrapolated as having an equal impact on a real clinical outcome (i.e. on a complex condition in a complex human body in a complex social environment).
...This builds up to tremendous distortions since private companies can simply not publish negative findings (the enforcement to pre-register clinical trials is abysmal), thus creating systemic publication bias!
b) Quick glance at logical fallacies (esp. regarding correlation vs. causation). A popular one is post hoc ergo propter hoc (if B follows A, then assume A caused B).
--Ex. Many people who get lung cancer drink alcohol, so drinking alcohol must cause lung cancer...? (Actually, many people who drink alcohol also smoke...)
c) The illusion of the shifting base: Ex. advertising a whooping 100% savings on $100... so it's free? Well, it turns out the savings was actually 50%. The "100% savings" is comparing the new (reduced) price of $50 to the savings ($50), when it should be comparing with the original base price of $100.
...This can get really sneaky the other way around, ex. locked into paying interest routinely on the original loan amount whereas you should pay on the new decreasing amount (since you've been paying off your loan)!
d) Plenty of counter-intuitive tricks can be done with percentages. Goldacre recommends using real numbers (1 out of 300) instead of percentages if you want to communicate more clearly/intuitively (esp. to describe changes). After all, an 100% increase could just be a change from 1-in-1,000,000 to 2-in-1,000,000.
The Missing:
--I have to add this note, because I’ve seen Bill Gates recommend this book (and novice conspiracy hacks getting the wrong ideas). Despite this book providing technical tools to spot manipulation, it presents “power” as an abstract bogeyman (mostly in the form of petty con artists).
1) Systemic power:
--Unlike the misleading narratives of petty con artists being “bad apples”, systemic power thrives in abstraction; this is how it builds social consent.
--To understand power, we must turn to political economy. Goldacre provides a useful bridge, by connecting these statistical deceptions to power in a step-by-step approach: starting with petty con artist quacks, then expanding to bigger systems: capitalist media and industries (including Big Pharma, which avoiding sensationalism which Big Media thrives on):
-Talking to My Daughter About the Economy: or, How Capitalism Works—and How It Fails
-Another Now: Dispatches from an Alternative Present
-Understanding Power: The Indispensable Chomsky
2) Propaganda:
--To expand our analysis of propaganda:
-Necessary Illusions: Thought Control in Democratic Societies
-Manufacturing Consent: The Political Economy of the Mass Media
-Merchants of Doubt: How a Handful of Scientists Obscured the Truth on Issues from Tobacco Smoke to Global Warming
November 11, 2009
It seems a little shallow to rate this semi-pamphlet at four stars, as one of the must-read books, but that's exactly what I'm going to do.
This book earns four stars from me simply from its concisiveness and practicality. You can churn through this beauty in one sitting. It is entertaining, has excellent examples, introduces concepts in a wry, witty tone, and after ten years of courses, articles, books, and opinions, I have yet to learn a single thing about misleading statistics that wasn't taught better and quicker in this book. You will sit down with this book for an hour or two and get up from your chair having a much more educated mind about the numbers that are constantly hurled at you.
I wish that there were one of these books for every topic imaginable to man. In my mind, the perfect library has a hundred thousand of these types of book. Short, simple, clear, and distinctly important.
September 2, 2018
کتاب، همانطور که نویسنده‌اش هم بیان کرده، ترجمه‌ای است از نسخه آمریکایی آن (نوشته دارل هاف-1954).
نویسنده اما اندکی در کتاب تصرف کرده و مثال‌هایی که بیشتر به درد ایران می‌خورد را آورده است. هرچند به گمانم این کار را نتوانسته آن طور که شایسته است انجام دهد!
کتاب، از نگاه من، چیز جدید و دندان‌گیری نداشت و با وجود اینکه نویسنده سعی دارد تفکر انتقادی را در خواننده رشد دهد اما این کار را نتوانسته آنطور که باید انجام دهد. مثال‌ها بسیار ساده و ابتدایی‌اند و اگر درس آمار را، در دبیرستان یا دانشگاه، پاس کرده باشید کتاب به هیچ‌وجه برایتان جذابیت و نکته تازه‌ای ندارد.
من اگر به جای نویسنده بودم؛ سعی می‌کردم مثالهای عینی‌تری را انتخاب کنم و یکی از بهترین مثال‌ها برای دروغگویی با آمار نگاهی به تاریخچه دولت‌ها است. نویسنده می‌توانست با نگاهی به مناظرات سیاسی ریاست‌جمهوری در ایران مثال‌هایی به مراتب ملموس‌تر و جالب‌تری انتخاب کند.
خیالتان هم راحت! تمام دولت‌ها، چه در ایران و چه در جهان، با آمار بازی کرده و می‌کنند و نویسنده می‌توانست حتی بی‌طرفی را حفظ کند و از همه جناح‌ها مثال بزند.
نویسنده اما اندکی در کتاب تصرف کرده و مثال‌هایی که بیشتر به درد ایران می‌خورد را آورده است. هرچند به گمانم این کار را نتوانسته آن طور که شایسته است انجام دهد!
کتاب، از نگاه من، چیز جدید و دندان‌گیری نداشت و با وجود اینکه نویسنده سعی دارد تفکر انتقادی را در خواننده رشد دهد اما این کار را نتوانسته آنطور که باید انجام دهد. مثال‌ها بسیار ساده و ابتدایی‌اند و اگر درس آمار را، در دبیرستان یا دانشگاه، پاس کرده باشید کتاب به هیچ‌وجه برایتان جذابیت و نکته تازه‌ای ندارد.
من اگر به جای نویسنده بودم؛ سعی می‌کردم مثالهای عینی‌تری را انتخاب کنم و یکی از بهترین مثال‌ها برای دروغگویی با آمار نگاهی به تاریخچه دولت‌ها است. نویسنده می‌توانست با نگاهی به مناظرات سیاسی ریاست‌جمهوری در ایران مثال‌هایی به مراتب ملموس‌تر و جالب‌تری انتخاب کند.
خیالتان هم راحت! تمام دولت‌ها، چه در ایران و چه در جهان، با آمار بازی کرده و می‌کنند و نویسنده می‌توانست حتی بی‌طرفی را حفظ کند و از همه جناح‌ها مثال بزند.
June 5, 2008
Yes, it has all the stuff you hear about: how people use stats to subtly (and not-so-subtly) misdirect the reader/listener, how to systematically recognize (or create) misinterpretations, and a strong implicit call to action for clearer information in public discourse.
But in the billion years since this classic came of age, we've all learned that other ways, some of them better presented. When it was written, many people believed the information they received in the papers, in magazines, and on the news. Now, news shows spend their time trying to discredit bloggers who point out their biases. Our cynicism has evolved to the point where How to Lie With Statistics teaches valuable technique, but loses much of its insight-producing novelty.
You should still read it, though, for two reasons:
- It's a classic. It's a great, simple read and you want to be able to say, "As it says in so classic and simple a book as How to Lie With Statistics--which, Professor, you have obviously studied--you are clearly hiding the truth!"
- No other book presents such a concise set of instructions for noticing when you have misled someone inadvertently. I frequently notice some document I'm preparing using a technique--quite often one built-in to popular business communication tools--that misleads people as to the real meaning of the data.
Because I've read this, I can catch myself and make sure I present my case clearly, but unimpeachably. If I mislead my audience, they'll catch me; They'll catch me and tear me apart, even if I were right.
So check out this classic, overlook its implicit innocence, learn some dirty tricks you may have forgotten or not caught, and pay attention to how you've been trained to use them just like we all have.
Bonus bit: my favorite bad statistics technique: Bar graphs with images for bars. As they grow taller, they grow wider, making a number twice as big appear four times as large.
But in the billion years since this classic came of age, we've all learned that other ways, some of them better presented. When it was written, many people believed the information they received in the papers, in magazines, and on the news. Now, news shows spend their time trying to discredit bloggers who point out their biases. Our cynicism has evolved to the point where How to Lie With Statistics teaches valuable technique, but loses much of its insight-producing novelty.
You should still read it, though, for two reasons:
- It's a classic. It's a great, simple read and you want to be able to say, "As it says in so classic and simple a book as How to Lie With Statistics--which, Professor, you have obviously studied--you are clearly hiding the truth!"
- No other book presents such a concise set of instructions for noticing when you have misled someone inadvertently. I frequently notice some document I'm preparing using a technique--quite often one built-in to popular business communication tools--that misleads people as to the real meaning of the data.
Because I've read this, I can catch myself and make sure I present my case clearly, but unimpeachably. If I mislead my audience, they'll catch me; They'll catch me and tear me apart, even if I were right.
So check out this classic, overlook its implicit innocence, learn some dirty tricks you may have forgotten or not caught, and pay attention to how you've been trained to use them just like we all have.
Bonus bit: my favorite bad statistics technique: Bar graphs with images for bars. As they grow taller, they grow wider, making a number twice as big appear four times as large.
May 17, 2021
A clever, creative way to teach how statistics manipulate every day life. (It made numbers approachable!) The original publishing date of 1954 seriously adds to the book's charm. The numbers he uses sound ludicrous today, but also underscore his points.
Definitely enjoyed this one.
Definitely enjoyed this one.
May 27, 2016
ایده اصلی کتاب ترجمه یک کتاب آمریکایی به همین نام از دارل هاف چاپ 1954است. منتها نویسنده بومی سازی و به روز رسانی کرده و چند تا مثال ایرانی هم آورده و به اصطلاح ترجمه و تالیف کرده... کتاب نکته ی خاصی ندارد. نکات ابهام برانگیز آمار مثل جامعه آماری و ابزارهای آماری (میانگین و واریانس و میانه) و رگرسیون و بیان همبستگی و نه بیان علت ومعلولی و بازی با نمودارها را که اهل کار با آنها بازی می کنند و حقایق را وارونه جلوه می دهند تشریح کرده... کتاب آمار و مدلسازی کتاب مشترک سه رشته ی انسانی و تجربی و ریاضی در سال های دبیرستان است. ولی جالبش این است که این کتاب آن قدر خشک است که قریب به اتفاق دیپلمه های ایران از آمار درکی ندارند. من اگر معلم آمار دبیرستان بودم در کنار آن کتاب خشک و بی خاصیت، حتما نفری یک نسخه ازین کتاب را هم به بچه ها می دادم. البته مثال های کتاب دقیق نیستند و مثال های ایرانی خیلی خیلی بهتر و دقیقتری می توانست وارد این کتاب شود. نویسنده به بهانه ی این که مثال های کتاب آمریکایی قدیمی است خودش را وارد ماجرا کرده. ولی در عمل می بینیم که همان مثال های کتاب 70 سال پیش خیلی دقیقتر و مشروح تر از مثالهای امروزی وطنی هستند... باری کاچی به از هیچی.
July 16, 2017
Noting that this book was published in 1954, one may instantly discount the information as outdated. However, there are recent events that can be related to some of the examples author Darrell Huff provides, and helps to increase the book's value.
For those who have fleetingly or never studied statistics, this is a good place to start. It is a quick and easily understandable read, written in plain English and with plenty of examples to prove the author's points. Personally, I have studied statistics (use and misuse) in various jobs, and have seen the positive as well as the detrimental aspects. Even with my background, I still found items of interest and was able to correlate some of Mr. Huff's thoughts to present day use.
The author's final chapter goes further than the explanations of the previous pages and outlines what the average person can do to avoid being fooled by deceptive statistics. The entire book is fun to read and informative. Four stars.
For those who have fleetingly or never studied statistics, this is a good place to start. It is a quick and easily understandable read, written in plain English and with plenty of examples to prove the author's points. Personally, I have studied statistics (use and misuse) in various jobs, and have seen the positive as well as the detrimental aspects. Even with my background, I still found items of interest and was able to correlate some of Mr. Huff's thoughts to present day use.
The author's final chapter goes further than the explanations of the previous pages and outlines what the average person can do to avoid being fooled by deceptive statistics. The entire book is fun to read and informative. Four stars.
February 25, 2017
A really fast read. And a fascinating one. Although I didn't pay attention to the release date before I began. So now I want to read another book discussing the same subject :3
These days, every claim is accompanied by stats to validate them. And when contradicting claims both have supporting studies behind them, you really have to stop and ask yourself what the fuck is going on. This is where this book comes to the rescue. Statisticians don't "lie" per se. But they do a lot of manipulation to bend the truth the way they want. So it's up for the reader to know how to bust them in the act, if we ever really wanted to make informed actions.
These days, every claim is accompanied by stats to validate them. And when contradicting claims both have supporting studies behind them, you really have to stop and ask yourself what the fuck is going on. This is where this book comes to the rescue. Statisticians don't "lie" per se. But they do a lot of manipulation to bend the truth the way they want. So it's up for the reader to know how to bust them in the act, if we ever really wanted to make informed actions.
March 15, 2010
أحب الرياضيات و تستهويني الأرقام ، لكن علم الإحصاء كان حاجزا لم أكن أستطيع تجاوزه ، و بالأخص ثلاثة مفاهيم (عقدتني) :
mean, mode, & medium
كنت دائما أقرأها في الدراسات التي أطلع عليها دون أن أفهمها أبدا ، و أشتهي لو أني أستطيع توظيفها في أبحاثي التي تعتمد على الأرقام .
فكّ هذا الكتاب عقدتي ! ، أخيرا فهمت ما تعنيه هذه المفاهيم من خلال تخطيط رسمي مبسط (<- بديل منزلي للسكانر العطلانة) ، لم أجده في أي من كتب الإحصاء !. متعة أني أخيرا عرفت معنى هذه المصطلحات تفوق الكثير من المتع !. سعيدة جدا بهذا (الاكتشاف) !.
الرسومات المزودة فيه مفيدة جدا و ساعدتني كثيرا في فهم المفاهيم . منها أيضا الطريقة التي استخدمها لتمثيل الخداع في الرسم البياني ، لكي تتضح الصورة للقارئ ، قام أولا بتقديم الرسم البياني الكامل الصافي ، و وضع في الخلفية صورة رجل مرسومة . و من بعدها ، أصبح التلاعب بالرسم واضحا جدا لأنه بنفس معدل التلاعب بالرسم البياني يظهر الخلل في الرسم للرجل في الخلفية . كان بديعا !.
تعلمت أيضا أن هذه المفاهيم الثلاثة قد تجتمع معا في الرسم الإحصائي حين يكون الرسم على شكل جرس ، و بالتالي استخدام أي منها سيكون ممثلا مناسبا عن المادة الإحصائية . لكن ما عدا ذلك ، لكل مفهوم قيمته الخاصة التي تحكي شيئا عن المادة .
و تعلمت ما هو الـ probable errors . هو معدل نسبي لنسبة الخطأ في الإحصاء الذي نقدمه ، بمعنى لنفرض أن إحصائياتي تتحدث عن عدد المرات التي يبكي بها عشر أطفال في الساعة على مدار اليوم . حين أراجع إحصائياتي ، لنفرض أني وجدت أن بمعدل كل طفل ، كنت أخطأ في التعداد بما يقارب معدل 3 مرات ، تزيد أو تنقص . 3 هو الـ probable error .
و في آخر فصل ، زود القارئ بعدد من الأسئلة التي يحتاج أن يسألها ليعرف مصداقية الإحصائيات المقدمة . هذه الأسئلة هي :
1. من الذي قال ؟ (هل للقائل يد في تشكيل النتائج بالصورة التي تظهر عليها ؟ أو هل له فائدة مجنية من ظهورها بتلك الصورة ؟ )
2. كيف توصل لهذه الأرقام ؟
3. ما الناقص في الإحصائيات المزودة ؟
4. هل استخدم الإحصائيات عن أ لاستنتاج شيء عن ب ؟ (غير الموضوع و استدلال في غير محله)
5. هل الإحصائيات كما تبدو منطقية ؟
اقتبست:
"Averages and relationships and trends and graphs are not always what they seem. There may be more in them than meets the eye, and there may be a good deal less."
mean, mode, & medium
كنت دائما أقرأها في الدراسات التي أطلع عليها دون أن أفهمها أبدا ، و أشتهي لو أني أستطيع توظيفها في أبحاثي التي تعتمد على الأرقام .
فكّ هذا الكتاب عقدتي ! ، أخيرا فهمت ما تعنيه هذه المفاهيم من خلال تخطيط رسمي مبسط (<- بديل منزلي للسكانر العطلانة) ، لم أجده في أي من كتب الإحصاء !. متعة أني أخيرا عرفت معنى هذه المصطلحات تفوق الكثير من المتع !. سعيدة جدا بهذا (الاكتشاف) !.
الرسومات المزودة فيه مفيدة جدا و ساعدتني كثيرا في فهم المفاهيم . منها أيضا الطريقة التي استخدمها لتمثيل الخداع في الرسم البياني ، لكي تتضح الصورة للقارئ ، قام أولا بتقديم الرسم البياني الكامل الصافي ، و وضع في الخلفية صورة رجل مرسومة . و من بعدها ، أصبح التلاعب بالرسم واضحا جدا لأنه بنفس معدل التلاعب بالرسم البياني يظهر الخلل في الرسم للرجل في الخلفية . كان بديعا !.
تعلمت أيضا أن هذه المفاهيم الثلاثة قد تجتمع معا في الرسم الإحصائي حين يكون الرسم على شكل جرس ، و بالتالي استخدام أي منها سيكون ممثلا مناسبا عن المادة الإحصائية . لكن ما عدا ذلك ، لكل مفهوم قيمته الخاصة التي تحكي شيئا عن المادة .
و تعلمت ما هو الـ probable errors . هو معدل نسبي لنسبة الخطأ في الإحصاء الذي نقدمه ، بمعنى لنفرض أن إحصائياتي تتحدث عن عدد المرات التي يبكي بها عشر أطفال في الساعة على مدار اليوم . حين أراجع إحصائياتي ، لنفرض أني وجدت أن بمعدل كل طفل ، كنت أخطأ في التعداد بما يقارب معدل 3 مرات ، تزيد أو تنقص . 3 هو الـ probable error .
و في آخر فصل ، زود القارئ بعدد من الأسئلة التي يحتاج أن يسألها ليعرف مصداقية الإحصائيات المقدمة . هذه الأسئلة هي :
1. من الذي قال ؟ (هل للقائل يد في تشكيل النتائج بالصورة التي تظهر عليها ؟ أو هل له فائدة مجنية من ظهورها بتلك الصورة ؟ )
2. كيف توصل لهذه الأرقام ؟
3. ما الناقص في الإحصائيات المزودة ؟
4. هل استخدم الإحصائيات عن أ لاستنتاج شيء عن ب ؟ (غير الموضوع و استدلال في غير محله)
5. هل الإحصائيات كما تبدو منطقية ؟
اقتبست:
"Averages and relationships and trends and graphs are not always what they seem. There may be more in them than meets the eye, and there may be a good deal less."
August 30, 2020
Very nice book on the most common statistical illusions present when loose statistics are presented in the media. It - statistical fooleries - still is surprisingly common and many examples can be observed in plenty of political campaigns and news outlets. It doesn't require a very deep knowledge of maths or statistics, so it is ideal if you are just looking to get a useful intuition on how popular statistical reporting typically works and where it fails.
November 22, 2017
I didn’t realize at first that this book was written in 1954. It’s still relevant today; math and people don’t change. It’s written in a fun, conversational style with lots of concrete examples that make the topics easy to understand. Even if you’ve already studied statistics, it’s a good refresher to see how they’re used in everyday media.
February 6, 2017
Το βιβλίο το αγόρασα πριν δυόμιση χρόνια με δυο ευρώ, σχεδόν στην τύχη. Μου είχε φανεί ενδιαφέρον το θέμα και χάρη στην τιμή δεν το σκέφτηκα παραπάνω. Να που ήρθε η ώρα να το διαβάσω κιόλας. Και πόσο χαίρομαι που τελικά το αγόρασα εκείνη την ημέρα και που το διάβασα τώρα.
Πρόκειται για ένα πολύ ωραίο, καλογραμμένο και εξαιρετικά ενδιαφέρον βιβλίο, που εξετάζει τις στατιστικές μεθόδους και έρευνες που χρησιμοποιούνται κατά κόρον από κυβερνήσεις, επιχειρήσεις, δημοσιογράφους, αναλυτές και διαφημιστές για να π��ίσουν το κοινό για την αλήθεια των λεγομένων τους. Το θέμα είναι ότι αυτά τα στατιστικά πολλές φορές είναι αναληθή, λάθος ή και παραπλανητικά. Ο συγγραφέας μυεί τον αρχάριο αναγνώστη στην τέχνη της στατιστικής παραπλάνησης και απάτης, μας δείχνει πως γίνονται τα στατιστικά λάθη αλλά και οι διάφορες απάτες και παραπλανήσεις. Το κάνει αυτό χρησιμοποιώντας πραγματικά περιστατικά παραπλάνησης του κοινού μέσω της στατιστικής, από άρθρα σε εφημερίδες και περιοδικά, μέχρι από έρευνες κυβερνητικών υπηρεσιών και ιδιωτικών επιχειρήσεων.
Όσον αφορά την όλη παρουσίαση, μου φάνηκε ιδιαίτερα προσιτή και κατανοητή. Η γραφή του Ντάρελ Χαφ είναι απλή και ευκολοδιάβαστη, διανθισμένη με χιούμορ σε ορισμένα σημεία. Ο συγγραφέας χρησιμοποιεί λογικά και πειστικά επιχειρήματα, με τα δεκάδες αληθινά περιστατικά να επιβεβαιώνουν τις θέσεις του. Διαβάζοντας με προσοχή αυτό το βιβλίο, ο αναγνώστης ίσως θα μπορεί να είναι σε θέση να βλέπει πίσω από τα διαγράμματα και τις στατιστικές μελέτες και έρευνες, να καταλαβαίνει πότε κάποιος του λέει ψέματα ή μισές αλήθειες, χρησιμοποιώντας στατιστικά δεδομένα. Ειδικά στην εποχή που ζούμε, κάτι τέτοιο είναι σίγουρα πολύ σημαντικό και ουσιώδες.
Φυσικά πρόκειται για ένα βιβλίο που πρέπει να ανατρέχει κανείς ξανά και ξανά, έτσι ώστε να εμπεδώσει πλήρως τους τρόπους εξαπάτησης και παραπλάνησης μέσω της στατιστικής. Άλλωστε, είναι ένα ευχάριστο και προσιτό οικονομικής/μαθηματικής φύσεως βιβλίο που μπορεί να διαβάσει και να απολαύσει κανείς. Η ελληνική έκδοση (Οξύ, 1997) σίγουρα ικανοποιητική, όμως εδώ και χρόνια εξαντλημένη...
Πρόκειται για ένα πολύ ωραίο, καλογραμμένο και εξαιρετικά ενδιαφέρον βιβλίο, που εξετάζει τις στατιστικές μεθόδους και έρευνες που χρησιμοποιούνται κατά κόρον από κυβερνήσεις, επιχειρήσεις, δημοσιογράφους, αναλυτές και διαφημιστές για να π��ίσουν το κοινό για την αλήθεια των λεγομένων τους. Το θέμα είναι ότι αυτά τα στατιστικά πολλές φορές είναι αναληθή, λάθος ή και παραπλανητικά. Ο συγγραφέας μυεί τον αρχάριο αναγνώστη στην τέχνη της στατιστικής παραπλάνησης και απάτης, μας δείχνει πως γίνονται τα στατιστικά λάθη αλλά και οι διάφορες απάτες και παραπλανήσεις. Το κάνει αυτό χρησιμοποιώντας πραγματικά περιστατικά παραπλάνησης του κοινού μέσω της στατιστικής, από άρθρα σε εφημερίδες και περιοδικά, μέχρι από έρευνες κυβερνητικών υπηρεσιών και ιδιωτικών επιχειρήσεων.
Όσον αφορά την όλη παρουσίαση, μου φάνηκε ιδιαίτερα προσιτή και κατανοητή. Η γραφή του Ντάρελ Χαφ είναι απλή και ευκολοδιάβαστη, διανθισμένη με χιούμορ σε ορισμένα σημεία. Ο συγγραφέας χρησιμοποιεί λογικά και πειστικά επιχειρήματα, με τα δεκάδες αληθινά περιστατικά να επιβεβαιώνουν τις θέσεις του. Διαβάζοντας με προσοχή αυτό το βιβλίο, ο αναγνώστης ίσως θα μπορεί να είναι σε θέση να βλέπει πίσω από τα διαγράμματα και τις στατιστικές μελέτες και έρευνες, να καταλαβαίνει πότε κάποιος του λέει ψέματα ή μισές αλήθειες, χρησιμοποιώντας στατιστικά δεδομένα. Ειδικά στην εποχή που ζούμε, κάτι τέτοιο είναι σίγουρα πολύ σημαντικό και ουσιώδες.
Φυσικά πρόκειται για ένα βιβλίο που πρέπει να ανατρέχει κανείς ξανά και ξανά, έτσι ώστε να εμπεδώσει πλήρως τους τρόπους εξαπάτησης και παραπλάνησης μέσω της στατιστικής. Άλλωστε, είναι ένα ευχάριστο και προσιτό οικονομικής/μαθηματικής φύσεως βιβλίο που μπορεί να διαβάσει και να απολαύσει κανείς. Η ελληνική έκδοση (Οξύ, 1997) σίγουρα ικανοποιητική, όμως εδώ και χρόνια εξαντλημένη...
August 12, 2017
In class 5 or 6, when we first started doing maths of finding average, mean or mode, I really had no idea what they meant or even why I was doing them. Just sum up the numbers, divide by their number and get the average or arrange them in increasing order, take the number in the middle - what they meant even. Maybe I was just following the instructions because the books said so or doing things this way will bring me marks in the exam. That's it. No more thoughts.
Later when we learnt regression analysis, even then I still didn't know why I am learning this. I saw teacher writing some x & y values on the board , then their was an equation y=ax+b. There were formulas for finding a and b. I just calculated the values of a and b. Later on if it was asked in the question, I would plug in a new value of x into the equation and get a new y. That's it.
But finally after attending numerical methods, machine learning and pattern recognition courses and while doing my thesis I know why we need them and what they mean :3
Later when we learnt regression analysis, even then I still didn't know why I am learning this. I saw teacher writing some x & y values on the board , then their was an equation y=ax+b. There were formulas for finding a and b. I just calculated the values of a and b. Later on if it was asked in the question, I would plug in a new value of x into the equation and get a new y. That's it.
But finally after attending numerical methods, machine learning and pattern recognition courses and while doing my thesis I know why we need them and what they mean :3
January 30, 2018
Written over 60 years ago, this is still a highly relevant book that exposes the many flaws in statistics and how easy it is to manipulate findings. A short book that everyone should read.
February 5, 2017
"The secret language of statistics, so appealing in a fact-minded culture, is employed to sensationalize, inflate, confuse, and oversimplify. Statistical methods and statistical terms are necessary in reporting the mass data of social and economic trends, business conditions, 'opinion polls', the census. But without writers who use the words with honesty and understanding and readers who know what they mean, the result can only be semantic nonsense… This book is a sort of primer in ways to use statistics to deceive. It may seem altogether too much like a manual for swindlers. Perhaps I can justify it in the manner of the retired burglar whose published reminiscences amounted to a graduate course in how to pick a lock and muffle a footfall: The crooks already know these tricks; honest men must learn them in self-defense." (10-11)I never cared much for statistics; I slid through Methods & Statistics 1, 2, and 3 relatively untouched, until the first time I did my own research and had to analyze the data I had so diligently collected. That was the point at which statistical analyses became meaningful to me—I had a theory and I was looking to see whether there was any evidence for it. The data itself means nothing, of course. Numbers do not mean anything on their own. What you do with the numbers, how you expose (non-)relationships between them, is when things get interesting. And tricky—as this little book sets out to show.
Huff uses humor to show a variety of ways in which statistics may be – and examples of how they have been – misleading. The book is dated by now, basing its examples mostly on stuff from the '20s through to the '50s. This did not really bother me, however, nor did I feel it detracted from the main points—if anything, it makes the book feel quaint while at the same time highlighting the fact that little has changed in how statistics are (ab)used in contemporary society. If you have a background in statistics, How to Lie is unlikely to teach you anything new about them. However, it is still worth reading if only to underscore the need to pay attention to the pervasive presence and use of statistics.
I, as someone trained in psychology, like to think that they are mostly used for good. Much knowledge has surely been gained thanks to the insights that increasingly sophisticated statistical analyses have offered. But an analysis is only as good as the way in which it is conducted, and results only as good as the way in which they are conveyed.
Huff closes his book on a more serious note, abandoning the burglar-revealing-his-trade spiel with five questions to ask when encountering statistical information (but don't worry, the chapter is still titled How to Talk Back to Statistics). They are worth listing and remembering:
1] Who says so? Look for bias: who has an interest in the statistic?
2] How does he know? Watch out for a biased/limited sample.
3] What's missing? The absence of information, like the number of cases or which kind of average was used, can render a statistic virtually meaningless.
4] Did Somebody Change the Subject? Make sure that the conclusion follows from the type of raw data that was collected.
5] Does It Make Sense? Think about what the statistic is supposed to mean/tell you, and ask whether it makes sense.
February 28, 2020
It turns out that there are so many ways one could deliberately lie with statistics, whilst simultaneously giving an air of credibility to whatever crap they are purporting.
This book is both scary and highly entertaining. It's a quick but informative read based on real-life examples.
Here's one that was particularly illuminating (emphasis mine):
"Let us say that during a period in which race prejudice is growing you are employed to “prove” otherwise. It is not a difficult assignment. Set up a poll or, better yet, have the polling done for you by an organization of good reputation. Ask that usual cross section of the population if they think blacks have as good a chance as white people to get jobs. Repeat your polling at intervals so that you will have a trend to report.
Princeton’s Office of Public Opinion Research tested this question once. What turned up is interesting evidence that things, especially in opinion polls, are not always what they seem. Each person who was asked the question about jobs was also asked some questions designed to discover if he was strongly prejudiced against blacks. It turned out that people most strongly prejudiced were most likely to answer Yes to the question about job opportunities. (It worked out that about two-thirds of those who were sympathetic toward blacks did not think the black had as good a chance at a job as a white person did, and about two-thirds of those showing prejudice said that blacks were getting as good breaks as whites.) It was pretty evident that from this poll you would learn very little about employment conditions for blacks, although you might learn some interesting things about a man’s racial attitudes.
You can see, then, that if prejudice is mounting during your polling period you will get an increasing number of answers to the effect that blacks have as good a chance at jobs as whites. So you announce your results: Your poll shows that blacks are getting a fairer shake all the time.
The worse things get, the better your poll makes them look."
This book is both scary and highly entertaining. It's a quick but informative read based on real-life examples.
Here's one that was particularly illuminating (emphasis mine):
"Let us say that during a period in which race prejudice is growing you are employed to “prove” otherwise. It is not a difficult assignment. Set up a poll or, better yet, have the polling done for you by an organization of good reputation. Ask that usual cross section of the population if they think blacks have as good a chance as white people to get jobs. Repeat your polling at intervals so that you will have a trend to report.
Princeton’s Office of Public Opinion Research tested this question once. What turned up is interesting evidence that things, especially in opinion polls, are not always what they seem. Each person who was asked the question about jobs was also asked some questions designed to discover if he was strongly prejudiced against blacks. It turned out that people most strongly prejudiced were most likely to answer Yes to the question about job opportunities. (It worked out that about two-thirds of those who were sympathetic toward blacks did not think the black had as good a chance at a job as a white person did, and about two-thirds of those showing prejudice said that blacks were getting as good breaks as whites.) It was pretty evident that from this poll you would learn very little about employment conditions for blacks, although you might learn some interesting things about a man’s racial attitudes.
You can see, then, that if prejudice is mounting during your polling period you will get an increasing number of answers to the effect that blacks have as good a chance at jobs as whites. So you announce your results: Your poll shows that blacks are getting a fairer shake all the time.
The worse things get, the better your poll makes them look."
July 3, 2015
I still wonder why Trigonometry and Calculus are offered in high school, but Statistics is not. It's such a broad subject that is used in so many fields-even forgetting all of the numbers we read in magazines. I digress.
This book specifically focuses on the facts and figures that we see everyday, pretty much everywhere. I thought it was well written and extremely thorough, going from problems that happen during study collection, to the cherry picking and presentation of data itself. I had to grin when I noticed that some of the exact same graph and picture tricks illustrated in this book are still being used today in some of my latest issues of Time & BusinessWeek (watch out for the 3-D bar graph). Overall a great, short, and extremely easy to read book that presents a nice reminder to the darker side of stats.
This book specifically focuses on the facts and figures that we see everyday, pretty much everywhere. I thought it was well written and extremely thorough, going from problems that happen during study collection, to the cherry picking and presentation of data itself. I had to grin when I noticed that some of the exact same graph and picture tricks illustrated in this book are still being used today in some of my latest issues of Time & BusinessWeek (watch out for the 3-D bar graph). Overall a great, short, and extremely easy to read book that presents a nice reminder to the darker side of stats.
July 8, 2017
I love the title.
The content feels outdated. I think people lie with statistics much better today than when this was published. Yay, we're improving!
The content feels outdated. I think people lie with statistics much better today than when this was published. Yay, we're improving!
January 9, 2019
Sneaky bastards.
April 12, 2022
Kind of hard to understand at times because math is confusing but the humor was sarcastic like me so bonus star for that
November 10, 2022
How can you avoid getting fooled by statistics—not only the ones you read or hear but also the ones you generate yourself? This simple (but not simplistic) guide explains how. Writing style is clear and to the point yet elegant. However, this was first published in 1954, so be prepared for persistent sexism and some old-fashioned expressions ("revenuers," "the blacks").
October 23, 2007
I'm just going to quote the Amazon.com review:
"There is terror in numbers," writes Darrell Huff in How to Lie with Statistics. And nowhere does this terror translate to blind acceptance of authority more than in the slippery world of averages, correlations, graphs, and trends. Huff sought to break through "the daze that follows the collision of statistics with the human mind" with this slim volume, first published in 1954. The book remains relevant as a wake-up call for people unaccustomed to examining the endless flow of numbers pouring from Wall Street, Madison Avenue, and everywhere else someone has an axe to grind, a point to prove, or a product to sell. "The secret language of statistics, so appealing in a fact-minded culture, is employed to sensationalize, inflate, confuse, and oversimplify," warns Huff.
Although many of the examples used in the book are charmingly dated, the cautions are timeless. Statistics are rife with opportunities for misuse, from "gee-whiz graphs" that add nonexistent drama to trends, to "results" detached from their method and meaning, to statistics' ultimate bugaboo--faulty cause-and-effect reasoning. Huff's tone is tolerant and amused, but no-nonsense. Like a lecturing father, he expects you to learn something useful from the book, and start applying it every day. Never be a sucker again, he cries!
Even if you can't find a source of demonstrable bias, allow yourself some degree of skepticism about the results as long as there is a possibility of bias somewhere. There always is.
Read How to Lie with Statistics. Whether you encounter statistics at work, at school, or in advertising, you'll remember its simple lessons. Don't be terrorized by numbers, Huff implores. "The fact is that, despite its mathematical base, statistics is as much an art as it is a science." --Therese Littleton
This is book aimed at the honest soul who is merely trying to make sense of the stream of numbers and stats pouring in from all around.
This is another must read.
"There is terror in numbers," writes Darrell Huff in How to Lie with Statistics. And nowhere does this terror translate to blind acceptance of authority more than in the slippery world of averages, correlations, graphs, and trends. Huff sought to break through "the daze that follows the collision of statistics with the human mind" with this slim volume, first published in 1954. The book remains relevant as a wake-up call for people unaccustomed to examining the endless flow of numbers pouring from Wall Street, Madison Avenue, and everywhere else someone has an axe to grind, a point to prove, or a product to sell. "The secret language of statistics, so appealing in a fact-minded culture, is employed to sensationalize, inflate, confuse, and oversimplify," warns Huff.
Although many of the examples used in the book are charmingly dated, the cautions are timeless. Statistics are rife with opportunities for misuse, from "gee-whiz graphs" that add nonexistent drama to trends, to "results" detached from their method and meaning, to statistics' ultimate bugaboo--faulty cause-and-effect reasoning. Huff's tone is tolerant and amused, but no-nonsense. Like a lecturing father, he expects you to learn something useful from the book, and start applying it every day. Never be a sucker again, he cries!
Even if you can't find a source of demonstrable bias, allow yourself some degree of skepticism about the results as long as there is a possibility of bias somewhere. There always is.
Read How to Lie with Statistics. Whether you encounter statistics at work, at school, or in advertising, you'll remember its simple lessons. Don't be terrorized by numbers, Huff implores. "The fact is that, despite its mathematical base, statistics is as much an art as it is a science." --Therese Littleton
This is book aimed at the honest soul who is merely trying to make sense of the stream of numbers and stats pouring in from all around.
This is another must read.
June 7, 2011
Recommended by both Jamie S. Z. and my Statistical Foundations professor. Really engaging and common-spoken, eager to make us adroit critical thinkers of statistical information. The main problem, of course, is its age, which enthusiastically describes plush neighborhoods with an average income of $15,000 and the enormous profits of $42 a week. Still, it has the fervor to educate us because, as H.G. Wells once prophesied, "Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write."
That being the case, I'd love to require this book for my rhetoric class. Am I over-stepping my bounds? I'm supposed to be teaching critical thinking skills, right? and I'm not sure that they'll all be taking a statistics class. I feel like when I teach graphic design or computer research or even logic, I'm doing what the experts say I ought, but can I teach statistics, too?
That being the case, I'd love to require this book for my rhetoric class. Am I over-stepping my bounds? I'm supposed to be teaching critical thinking skills, right? and I'm not sure that they'll all be taking a statistics class. I feel like when I teach graphic design or computer research or even logic, I'm doing what the experts say I ought, but can I teach statistics, too?
October 4, 2013
OK, first off, it isn't normal that I give a math book 5 stars. I often find them dull, boring, and difficult to read. However, How to lie with statistics was as funny as it was informative. Duff does a good job of not only explaining what tricks people use on statistics to twist the facts, but he gives poignant examples that were just as relevant when he wrote this book as they are today. What I found most interesting is how he dissected the "logic" that uses these techniques to explain how they did it, what they trick your brain into seeing, and how you can question it effectively.
Duff talked about sample biases, which we can see in every day research. While his example was that of Yale graduates pay grades is a bit outdated, he shows how this sample represents a small number of people, and how it is most certainly a false representation based on logic, common sense, and science.
He then went on to dissect the differences between mean, median and mode, followed by inadequate samples (think sales pitch where you only ask 3 people versus more), tests that mislead and reveal nothing (like IQ tests), manipulation of graphs and scales, semi-attached figures, post hoc representations, and statistical manipulations. He did the same thing with each topic to show how you start with facts, manipulate it, and present it. Then he shows you how it's wool covering your eyes.
The last chapter was my favourite. It tells you how to pull the wool from your eyes and argue back against it by pointing out their chicanery. Essentially it was a recap of the previous chapters, then telling you not to be afraid to call out the BS you see in the media, advertising, or politics.
Duff does all this in a tongue-in-cheek way. He pulls no punches, nor does he sugar coat things. When I read this I could easily pick out several examples of every single thing he talked about from my own environment. Lies the media tells me to say so it's "politically correct", lies on the magazine back advert, graph manipulation in my precious National Geographic or Time Magazine... it was a disappointment to see how pervasive it is in our society. It's even more disappointing knowing that most people do not know what they are looking at and do not know that they are being swindled.
This is a book that everyone should read so they are more well informed about the world around them and can pull the wool from their eyes.
Duff talked about sample biases, which we can see in every day research. While his example was that of Yale graduates pay grades is a bit outdated, he shows how this sample represents a small number of people, and how it is most certainly a false representation based on logic, common sense, and science.
He then went on to dissect the differences between mean, median and mode, followed by inadequate samples (think sales pitch where you only ask 3 people versus more), tests that mislead and reveal nothing (like IQ tests), manipulation of graphs and scales, semi-attached figures, post hoc representations, and statistical manipulations. He did the same thing with each topic to show how you start with facts, manipulate it, and present it. Then he shows you how it's wool covering your eyes.
The last chapter was my favourite. It tells you how to pull the wool from your eyes and argue back against it by pointing out their chicanery. Essentially it was a recap of the previous chapters, then telling you not to be afraid to call out the BS you see in the media, advertising, or politics.
Duff does all this in a tongue-in-cheek way. He pulls no punches, nor does he sugar coat things. When I read this I could easily pick out several examples of every single thing he talked about from my own environment. Lies the media tells me to say so it's "politically correct", lies on the magazine back advert, graph manipulation in my precious National Geographic or Time Magazine... it was a disappointment to see how pervasive it is in our society. It's even more disappointing knowing that most people do not know what they are looking at and do not know that they are being swindled.
This is a book that everyone should read so they are more well informed about the world around them and can pull the wool from their eyes.
October 19, 2018
First few chapters made sense. Post that, I started questioning my own reasons for picking this up in the first place.
While it's a small book already, it's summary would have sufficed just as well.
While it's a small book already, it's summary would have sufficed just as well.
Displaying 1 - 30 of 1,655 reviews