How accurate can models from social media be?

Slides:



Advertisements
Similar presentations
Sampling Distributions Suppose I throw a dice times and count the number of times each face turns up: Each score has a similar frequency (uniform.
Advertisements

1 Introduction to Inference Confidence Intervals William P. Wattles, Ph.D. Psychology 302.
Predicting White Wine Quality Scores RAPHAEL MWANGI.
1 IQ and Personality Assessment Psychology 631 William P. Wattles, Ph. D.
Probability Probability; Sampling Distribution of Mean, Standard Error of the Mean; Representativeness of the Sample Mean.
REGRESSION What is Regression? What is the Regression Equation? What is the Least-Squares Solution? How is Regression Based on Correlation? What are the.
Basic Statistical Concepts Psych 231: Research Methods in Psychology.
Basic Statistical Concepts
Accuracy of Prediction How accurate are predictions based on a correlation?
Basic Statistical Concepts Part II Psych 231: Research Methods in Psychology.
REGRESSION Predict future scores on Y based on measured scores on X Predictions are based on a correlation from a sample where both X and Y were measured.
Measures of Variability: Range, Variance, and Standard Deviation
From Last week.
Data Collection & Processing Hand Grip Strength P textbook.
Technical Adequacy Session One Part Three.
Testing Theories: Three Reasons Why Data Might not Match the Theory Psych 437.
Estimation Statistics with Confidence. Estimation Before we collect our sample, we know:  -3z -2z -1z 0z 1z 2z 3z Repeated sampling sample means would.
Individual values of X Frequency How many individuals   Distribution of a population.
Simple Linear Regression One reason for assessing correlation is to identify a variable that could be used to predict another variable If that is your.
1 rules of engagement no computer or no power → no lesson no SPSS → no lesson no homework done → no lesson GE 5 Tutorial 5.
Part IV Significantly Different: Using Inferential Statistics
TYPES OF STATISTICAL METHODS USED IN PSYCHOLOGY Statistics.
Measures of Central Tendency And Spread Understand the terms mean, median, mode, range, standard deviation.
Plan and Data. Are you aware of concepts such as sample, population, sample distribution, population distribution, sampling variability?
IE(DS)1 Many of the measures that are of interest in psychology are distributed in the following manner: 1) the majority of scores are near the mean 2)
Chapter 4: Variability. Variability Provides a quantitative measure of the degree to which scores in a distribution are spread out or clustered together.
Week 6. Statistics etc. GRS LX 865 Topics in Linguistics.
Intro to Psychology Statistics Supplement. Descriptive Statistics: used to describe different aspects of numerical data; used only to describe the sample.
Education 793 Class Notes Inference and Hypothesis Testing Using the Normal Distribution 8 October 2003.
Variability. What Do We Mean by Variability?  Variability provides a quantitative measure of the degree to which scores in a distribution are spread.
Uncertainty and error in measurement. Error Uncertainty in a measurement Limit to the precision or accuracy Limit to the reliability An error is not a.
Intro to Statistics for the Behavioral Sciences PSYC 1900 Lecture 7: Regression.
Z-Scores. Histogram A bar chart of a frequency distribution. 0 — 2 1 — 3 2 — 0 3 — 2 4 — 4 5 — 3 6 — 3 7 — 5 8 — 3 9 — 2 10—2.
Chapter Eleven Sample Size Determination Chapter Eleven.
Data Screening. What is it? Data screening is very important to make sure you’ve met all your assumptions, outliers, and error problems. Each type of.
Statistics. “Oh, people can come up with statistics to prove anything. 14% of people know that” Homer Simpson.
Statistics for the Behavioral Sciences (5th ed.) Gravetter & Wallnau
9.3 Hypothesis Tests for Population Proportions
Statistical analysis.
Statistics for the Social Sciences
Statistical analysis.
Math in Science + Graphs
Item Analysis: Classical and Beyond
Simulation-Based Approach for Comparing Two Means
Research methods Lesson 2.
Quantitative Data Analysis P6 M4
Free teaching material for a fact-based worldview
Free teaching material for a fact-based worldview
Free teaching material for a fact-based worldview
Research Statistics Objective: Students will acquire knowledge related to research Statistics in order to identify how they are used to develop research.
PSY 626: Bayesian Statistics for Psychological Science
Correlations: testing linear relationships between two metric variables Lecture 18:
Standard Deviation.
Reasoning in Psychology Using Statistics
Arithmetic Mean This represents the most probable value of the measured variable. The more readings you take, the more accurate result you will get.
Intro to Psychological Testing (part II)
Ensembles.
Statistics II: An Overview of Statistics
Inferential Statistics
Statistics for the Social Sciences
Sampling distributions:
Estimating Population Parameters Based on a Sample
Significance Tests: The Basics
Item Analysis: Classical and Beyond
Topic 1 Statistical Analysis.
Quantitative design: Ungraded review questions
National Curriculum Assessments 2019
But… it is still your responsibility
Item Analysis: Classical and Beyond
STA 291 Spring 2008 Lecture 17 Dustin Lueker.
Presentation transcript:

How accurate can models from social media be?

Why is assessing personality accuracy based on Facebook data hard? Personality measured on an arbitrary scale Why is assessing personality accuracy based on Facebook data hard? Correlations are not intuitive

We can use age to build intuitions about accuracy Average age in USA: 41 years old Average SD in USA: 23.5 years old We can simulate age data and check accuracy at different correlation levels

How to assess accuracy well? Error Actual age Predicted age If Mark is 40 years old, and I predict he is 30, the error is 10 years. If Sally is 20 years old, and I predict she is 25, the error is 5 years. If Liz is 35 years old, and I predict she is 55, the error is 20 years. We can average this error across many people to see how accurate a prediction is on average.

How accurately can we predict age? Random guess Assume everyone is 41 On average, wrong by 27 years On average, wrong by 19 years Prediction with r = .3 On average, wrong by 18 years

But…predicted age with .3 looks unrealistic The predicted scores are clustered heavily around the average—not usable since you would conclude most people are middle aged So need to spread the scores out to look more like the true distribution Average error of spread out predictions is 22 years…worse than assuming everyone is average!

How accurately can we predict age if we can make more accurate models? R = 0.5 (quite hard to do) R = 0.8 (maximum realistic accuracy ever for personality) On average, wrong by 19 years On average, wrong by 12 years R = .95 (extremely unlikely to ever achieve this for anything psychological) On average, wrong by 6 years

Another way to see it…. No relationship Perfect relationship R = 0.3

Such predictions are useful for understanding the population, but mind control or actual knowledge of specific people is science fiction