W9 Flashcards
Content analysis
quantitative, systematic and objective technique for describing the manifest content of communication
used for any type of content, any recorded medium, press, radio, video, etc.
Advantages of quantitative content analysis
- emphasis on systematic sampling
- clear definitions of units and counting
- 100% reliability if the researcher uses computers because analysis will code all occurrences in the same way
Disadvantages of quantitative content analysis
- address only questions do content
- really useful just for comparisons which limits its applicability
- validity in content analysis can be problematic in terms of relating its findings to the external world
- e.g. the frequency with which the word patriotism appears in a politicians speech increased over time does not entitle us to assume that in fact, the politician has become more patriotic over time.
Data reduction
necessary to detect any patterns in the raw data and to make any such patterns comprehensible to readers of the final research report
Content analysis
conventional use -> analyzing media content such as news, entertainment or advertising
the appeal of the web content analysis is ready and inexpensive access to a huge diversity of content from a variety of media types and sources worldwide
Concerns with web sampling
- not fully knowing the relationship between the sample and the population from which it is drawn
- lack info about the source content of these data may be inaccessible
- samples may be bias for human reasons or technical reasons
- ability of software to recognize and analyze subtleties of human language and meaning for example, humor.
Computer analysis of content
simplified by:
1) stemming - changing all variations of a word to its basic stem; e.g. fish, fishing, fished, fisherman
2) lemmatization - grouping words together based on their basic dictionary definition so that they can be analyzed as a single item; e.g. car and automobile do not have a common stem but both describe the word vehicle
3) stop words - high frequency words such as pronouns and prepositions
Content analysis as quantitative and qualitative
scientific quantitative approach to the analysis of content, while allowing for qualitative subjective analyses
they coexist in any given study. one can use quantitative analysis to investigate qualitative phenomena such as emotional states
qualitative studies ultimately require qualitative interpretations.
non directional research hypothesis
reflects a difference between groups but the direction of the difference is not specified
directional research hypothesis
reflects a difference between groups and the direction of the difference is specified
null
no relationship
population
indirectly tested
written with =
research h
relationship
sample
directly tested
written with ><
normal curve
1) symmetrical
2) asymptotic tail
3) mean median and mode are equal
if median and mean are different the distribution will be skewed to the left or to the right
the asymptotic tails mean they come closer and closer to the horizontal axis but never touch
distance between mean and 1 stdev
34% of all cases under the curve
1 and 2 stdevs
14% of cases under the curve
2 and 3 stdevs
2% of cases under the curve
3 stdevs<x
0.1% of cases under the curve
z-score
standard score, comparable because they are standardized in units of stdevs
+ z scores = right of the mean, upper half
- z scores = left of the mean, lower half
84% fall below a z score of 1 (50% below the mean and 34% between mean and z score of 1)
16% of all scores fall above a z score of +1
skewness
measure lack of symmetry
positively skewed - tail on the right
negatively skewed - tail on the left
kurtosis
has to do with how flat or peaked a distribution appears
1) platykurtic - relatively flat distribution compared to a normal curve
2) leptokurtic - relatively peaked and taller distribution compared with a normal distribution
significance level
risk associated with not being 100% confident that you observe in an experiment is due to the treatment or what was being tested
risk willing to take to reject a null hypothesis when it is actually true
one sample Z TEST
comparison between a samples mean score and a populations mean score
utilized to compare a sample statistic with a population parameter to see if the sample is representative of the population
when to perform a t-test for the significance of correlation coefficient?
1) im examining relationships between variables
2) im dealing with 2 VARIABLES
when to perform regression, factor analysis or canonical analysis?
1) im examining relationships between variables
2) im dealing with MORE than 2 VARIABLES
when to perform dependent samples t-test?
1) im examining differences between groups on one or more variables
2) participants are being tested more than once
3) im dealing with 2 GROUPS
when to perform repeated measures ANOVA?
1) im examining differences between groups on one or more variables
2) participants are being tested more than once
3) im dealing with MORE than 2 GROUPS
when to perform independent samples t-test?
1) im examining differences between groups on one or more variables
2) participants are being tested ONCE
3) im dealing with 2 GROUPS
when to perform one way ANOVA?
1) im examining differences between groups on one or more variables
2) participate are being tested ONCE
3) im dealing with MORE than 2 GROUPS
effect size
strength of a relationship between variables (correlation coefficient or value that estimate differences)
relationship between variables can also be apparent in the size of a difference between groups
small effect size - 0-0.2
medium effect size - 0.2-0.8
large effect size - 0.8 or higher
inferential stats estimates
statistical inference: information that allows us to draw conclusions about the entire population from which we got our sample from
point estimate
single number, best guess for estimation parameter
- x bar (mean) is a good point estimate for parameter mu (mean)
- one individual point estimate is no a good reference to see If the estimate is close to the population parameter we want to see
interval estimate
range of values within which we expect the parameter to fall
probability that interval contains population value is called the confidence level
hypotheses testing
expectations about the parameters researchers are interested in - hypotheses key ingredients of significance testing
we assess whether a hypothesis makes sense or not through significance test
on a significance test you always assume NH is true unless proven otherwise
rejected if AH has enough proof to make H0 wrong (doesn’t mean H0 is true)