Intro Methods Exam Review final.docx
SchoolUniversity of Guelph
DepartmentSociology and Anthropology
Course CodeSOAN 2120
This preview shows pages 1-2. to view the full 6 pages of the document.
Intro Methods Exam Review
Purpose of course-An introduction to research methodology in the social sciences.
Quantitative vs. Qualitative-Quantitative tests theories, objective, free from bias, reliable, but is
less thorough and less valuable to sensitive issues.
-Qualitative creates theories, subjective, not free of bias, less reliable but more thorough.
-e.g. Evaluation for exams often starts off with mostly multiple choice with little long answer.
Easier to evaluate that way with large numbers of people, quicker to mark, same answers for
everyone but is it really an accurate way of measuring your knowledge? It then moves into more
of an essay format which is much more thorough and valid and we can do this because there are
less students however it still takes a long time to evaluate. But is everyone then measured by the
Experimental Designs-Controlled environment modeled after the natural sciences. 3stages,
Pretest (often irrelevant), Treatment (Apply the experiment to one thing/group), Posttest (results
we get that we can compare).
e.g. Theory-Drug reduces cancer cells
Hypothesis-Cancer cells exposed to the drug will deteriorate.
Pretest-measure size of cells
Treatment-Put some cells is 1 dish, more in another, apply treatment to one cell. Cover
dishes to prove nothing else has influenced the cells.
Posttest-Measure size of cells again and see that the treatment one has gone smaller.
Conclusion-does reduce cancer cells; controlled experiment.
Social experiement-60’s psychological experiment on feelings of happiness. Have 3 groups,
MPT, SC and control group.
Theory-MPT and SC increase happiness
Hypothesis-Groups 1 and 2 will be happier than group 3
Pretest- see how happy they feel in a 5min period
Treatment- Administer MPT, SC and water
Posttest- See how happy they were in a 5 min period.
Survey Research-Most research by sociologists and psychologists is surveys
-No manipulation of subjects, popular methods are by phone, mail, personal interview, internet
etc. Each person on the survey represents a variable.
-Advantages-Tap into attitudes, easy to administer to large numbers of people, very generalizable
to the population.
-Disadvantages-Lack of control, e.g. people are already in the groups you didn’t put them there.
Variable-Anything that varies; has more than one category. Ask a question and there is more
than 1 answer, the question is a variable.
Categorical Variable-No rank order, non-numerical. E.g. religion, gender etc.
Quantitative variable-Continuous (ranked) and numerical. E.g. income in dollar amounts.
Measures of central tendency-Statistics that summarize the central location of Quantitative
Mean-Add all numbers together then divide by the number of numbers
Median-Organize numbers in order from lowest to highest. Pick the middle number.
Mode-Most frequently occurring number.
Distribution curves-show the distribution of all numbers on a graph
Positively skewed-Most observations on the left, mean is greater than median.
Only pages 1-2 are available for preview. Some parts have been intentionally blurred.
Negatively skewed-Most observations on the right, mean is less than mean.
Normal Distribution-this is the bell curve, mean equals median. Important because it allows us to
calculate standard deviation.
Variance-Calculate the mean; first value-mean=x, then square that value. Once you have all he
squares add them together then divide this number by the sample size minus 1.
Standard Deviation-Square root both sides of the variance.
Validity-Is your measure measuring what you intended it to measure. If it is valid it is also
Reliability-Are our measurements consistent? Do we get the same assessment over again?
Reliability does not mean it is valid.
Selection Bias-If a convenience sample is used. E.g. use university students to asses BMI’s for
Guelph. It is skewed by the fact that only university students are studied. Tends to happen in
magazines. Random sampling usually fixes this but still has issues. However, statistics can be
adjusted to compensate for selection bias.
Statistics and Bias-Bias of a statistic is the difference between its average value (obtained from
the sampling distribution) and the true value of the parameter.
-A statistic is unbiased if the mean of its sampling distribution is equal to the parameter being
estimated. Only works if it is a random sample.
Sampling Error-sampling error is the difference between the statistic and the parameter due to
random processes. Sampling error decreases if it is a random sample and the larger the sample
Central Limit Theorem-Regardless of the population distribution with repeated sampling the
shape of the sampling distribution will be approximately normal with the population parameter at
its center. This theory is important for hypothesis testing and to help calculate z-scores for
Law of large numbers-If we repeat a random phenomenon many times the average value will
be closer to the population parameter. Larger samples is more likely a statistic will represent the
population. Size of the sample is more important that the size of the sample in relation to the size
of the population.
z-scores-Tells us how many standard deviations a value is from the mean which allows us to
calculate statistical significance. Things are statistically significant at + or -2sd from the mean
(0.05 that it could happen by accident) and at + or -2.6sd from the mean (0.01 that it could
happen by accident).
-to calculate z-score do a value minus the mean divided by the standard deviation
Calculating p-values-Use z-scores to guess that probability something happened by accident. +
or -2sd means 95% confident it didn’t happen by accident. + or -2.6 means 99% confident it
didn’t happen by accident.
Hypothesis (significance) testing-Tests claims about the population (if the mean of the
population is a specified value and for the difference between the two values). Tests are based on
the sampling distribution; significance tests are based on z-scores.
Step 1-State the null hypothesis which is 0 (we are testing this). Ho=0
You're Reading a Preview
Unlock to view full version