Methods for Data Analysis
Content analysis chapter 29
primary focus recorded information and people's relationship with it/focuses on features of recorded information – adopted as a useful information and library science science research technique
Use of content analysis
describes a message pool – identify relationships between message characteristics
infer or predict outcomes or effects of messages/characteristics of message content link to measures of human response/other measures
manifest and latent content
manifest observable and countable
latent content difficult/impossible to count – level of anxiety for example
content analysis concerned only with manifest content – validity may be questionable
Units of analysis
sampling units/recording units
sample from the overall population
recording units – elements of content that are coded, physical, conceptual, temporal
straightforward or not
Sampling senses – all members included
content analysis – obtain a representative sample which can be generalized to the population
Coding scheme development –instruction book and data collection instrument for use in conducting content analysis
first step identify the critical variables you wish to examine/grounded in prior research and theory step two choose or define indicators for those variables/what you will count and code for, often manifest content indicated presence of latent variables – require attention to validity reliability accuracy precision
indicators and content analysis take two forms definition of content characteristics (features it can be counted) and sets of categories or levels to record codes presenting the content of the message
Categories must be exhaustive – cannot be satisfied by other not present cannot determine
categories much be must be mutually exclusive – no recording should belong in more than one category of a particular variable
Multiple coders and inter-coder agreement
strives for objectivity and replicability
given same content, coding scheme and training, and he coders work should result in the same data. High measures of intercoder agreement indicate reliability of the results of the coding process
Analysis of coded data – numerical data to analyze
Computer supported content analysis
may not be appropriate for every content analysis/some better suited to computer supported analysis
Chapter 30 qualitative analysis of content
Content analysis primarily used in ILS as quantitative research method until recently/current studies use qualitative content analysis defined as
research method for the subjective interpretation of the content of text data through the systematic classification process of coding identifying themes or patterns
An approach of empirical methodical controlled analysis of text within the context of communication, following content analytic rules and step-by-step models, without rash quantification
any qualitative data reduction and sense making effort that takes a volume a qualitative material and attempts to identify core consistencies and meanings
Inductive versus deductive
qualitative content analysis – condense raw data into categories or themes based on valid inference and interpretation new paragraph three approaches to qualitative content analysis
conventional qualitative content analysis – coding categories derived directly in and deductively from the raw data
directed content analysis – coding starts with a theory or relevant research findings, researchers immerse themselves in the data and allow things to emerge
summative content analysis – starts with counting of words or manifest content/extensive analysis to include latent meanings and themes
Process of qualitative content analysis move back and forth between concept development and data collection
Involves a set of said systematic and transparent procedures for processing data
step one prepare the data
step to define the unit of analysis
step three develop categories in the coding scheme
step four tester coding scheme on a sample of text
step five code all the text
step six assess your coding consistency
step seven draw conclusions from the coded data
step eight reports your methods and findings
Computer support for qualitative content analysis
supported by computer programs such as NV IVO, AATLAS.TI
common purpose – assist researchers in organizing managing encoding qualitative data in an efficient manner
text editing, note and memo taking, coding, text retrieval and node/category manipulation
in corpse visual presentation module – see relationships between categories more vividly
recorded coding history to allow researchers to keep track of evolution of their interpretations
Trustworthiness
validity reliability and objective is to be our used to evaluate the quality of research in the conventional cost of the test research paradigm
credibility
adequate representation of constructions of the social world understudy
transferability – extent to which the researchers working hypothesis can be applied to another context
dependability – coherence of the internal process and the way the researcher calls for changing conditions of the phenomena
Qantas qualitative content analysis is a valuable alternative to more traditional quantitative content analysis – working in interpretive paradigm
identify important themes or categories within a body of content and provide a rich description
Chapter 31 discourse analysis
used to uncover other meanings that we negotiate in our everyday in professional interactions but are rarely made explicit within those interactions – analysis of discourse
Methods and tools – no analytic method at least is a term is understood elsewhere in social psychology – broad theoretical framework – attention on the constructive and functional dimensions of discourse
Instead of using scales or counting occurrences includes the works of theorists within ILS – suggest a number states there will be taken during discourse analysis (not set in stone/sometimes overlapping/completely absent)
First step construct your research question
step two select a sample of the course to study
next step collect records and documents
ready to code data – identifying themes within categories of emerge and take shape as you examine the tax
success predicted on close reading and rereading of the text
looking for nuance contradictions areas of vagueness or ambivalence next
final step validate your findings
Discourse analysis – strong proof/potential to explore passive inquiry to provide greater understanding/required rigor clarity and self inquiry
Chapter 32 analytic induction
induction is one of the main forms of scientific logic, best understood in contrast with other main form of scientific logic: deduction
deductive reasoning – logical process of arriving at specific facts it must be true given a broader sense that seem to be true
induction – logical process of arriving at a general conclusion by a examining set of specific backs
analytic induction – specific form of inductive reasoning to analyze qualitative data
Process of analytic induction
first formulate a rough description of the phenomena you want to explain
next develop a hypothetical explanation of the phenomena
derive your initial definition and hypothesis continue choosing cases and studying them to see if emergency of information is consistent with hypothesis
examination of cases
Challenges in using analytic induction
when to stop examining more cases/no were all negative cases may be lurking
analytic induction has been used relatively infrequently especially in ILS
Analytic induction is an analysis methods not been used in ILS studies yet – holds promise for application to research questions/often working with qualitative data
Chapter 33 descriptive statistics
describing a phenomena with statistics
variables and their levels of measurement
measures of central tendency
mean
median
mode
Choose which method of central tendency to use
Measures of dispersion
range and interquartile range
variance and standard deviation
confidence intervals
Summarize with simple statistics – include at least one measure of central tendency and one manner measure of dispersion to interpret findings to draw valid conclusions from the data
Friday basic charts and graphs for frequency distributions
Descriptive statistics are the most essential view of your study findings critical components of any report of any research
for each variable of interest you should report with a measure of dig central tendency (mean median or mode)
measure of dispersion (standard deviation, range or interquartile range)
starting point for understanding your findings
Chapter 34 frequencies cross tabulation and the chai – square statistic
frequency distributions
counts can be organized and displayed in a table referred to as the frequency distribution
two-way frequency distributions: cross tabulation tables – displays frequency of categories of a single variable
chai square statistic – test with the frequency distribution is likely to occur by chance
Visualizations of your data set
pie charts
bar charts
histograms
line graphs
box plots
lying with graphs
Worthwhile to report descriptive statistics on his findings included frequency distributions on key category of variables
reported in tables – frequency and relative frequency in each cell
row and column totals should be displayed
chai square statistics used to test statistical significance possible relationship between variables in a two-way table
graphical display of the frequency
clearly understand the study results and communicate to the audience
Chapter 35 analyzing sequence he of events
behaviors happen over time
coding the data
approaches to sequential event analysis
Mark of models: examining sequences step by step
optimal matching approaches: comparing whole sequences
visualizing sequences of events
Advantages and event to disadvantages to each – may go beyond your level of expertise
before making a decision about which approach to use in your study, examine that analysis of sequences of search moves
Sequence of events, not enough to know which occur most frequently – need to know the sequence of those actions and which sequences occur most frequently
step-by-step approach using simple first order Markov models and transition matrix, look for frequently occurring longer sequences, pattern analysis, similarity between sequences through optimal matching algorithms
Chapter 36 correlation
statistical analysis method that helps to examine the relationship between two variables
proportion of the variability in one variable explained in the variability of the other variable
if perfectly correlated explains variability
two variables not correlated – none of one is explained by the other
linearly really related and data are ordinal interval or ratio data
Selecting the appropriate correlation statistic
some cautions for interpreting the correlation coefficient
Many research questions are oriented to discovering whether there is a relationship between two variables, how strong the relationship is
whether one variable in some cases causes the other
correlation cannot directly address – provide evidence of a relationship between the two
Chapter 37 comparing means: T tests and analysis of variance
compare different groups in terms of the mean ( average)
two statistical techniques – the t-test and an alley analysis of variance ANOBA
comparing two groups the t-test – probability a difference between two means
interpreting the results of the t-test
statistically significant results
nonsignificant results
Comparing multiple groups
more than two groups to compare
interpreting the results
Many research questions in information and library science of all making comparisons – comparing the effectiveness/performance
calculated amine for each group or entity you want to compare
variability across different samples and within each sample
likelihood indifference hold true across entire population
examine the size of the difference
Chapter 38 putting it all together conclusion
starting over from the beginning
some examples of how study can be put together financial winding your way through this book
providing evidence to promote progress in chapter information and library science
No comments:
Post a Comment