Analysis & Interpretation of Data-stat ppt

statistics

1. Dr.Geethakumary V.P

2. PLAN FOR DATA ANALYSIS
Consider
Research question
Objectives
Variables
Hypotheses and suitable tests, how to
interpret
Available software

3. Data management is easy
Definite plan quickens the data analysis
process
Adequate plan enables to prepare early
for analysis
Increases the scientific integrity

4. PHASES/ STEPS
IPre-analysis
IIPreliminary assessment
IIIPreliminary action
IV Analysis
VInterpretation

5. Login, Check
Edit raw data
Select software package
SPSS - Statistical package for social sciences
SAS - Statistical analysis system
NCSS - Number crunch statistical system
CRUNCH
ABSTAT
BMDP- Biomedical data processing
Contd..

6. Code data
Pre – Categorized
Un – Categorized
Coding missing value
Enter into computer file

Contd..

7. Inspect data for

Outliers (values outside normal range)

Wild codes (not possible)

Irregularities – inconsistent
Clean data

Check visually

Enter data twice
Create & document analysis file

8. Assess missing values

Extend of the problem

Cumulative extend of variables
frequency

Randomness of missing values (use
split half method)

Estimate value by use of SPSS
package

9. Assessing quality of data
Internal consistency
Limited variability
Extreme skewness
Assessing bias
Reliability
Validity

10. Delete missing cases (as per extend of the
problem)
Delete variable
Estimate mean of variable
Substitute by mean/median value specially in
‘scales’
Use multiple regression

11. IV DATA ANALYSIS
Measurement
It involves assigning numbers & objects
Measurement errors include
Situational contaminants
Response set bias
Transitory personal factor
Fatigue
Tension, etc.
The normal curve
Theoretical normal curve
Central limit theorem
Sampling distribution
Standard deviation
It measures sampling error level of significance
Clinical significance
Descriptive and explorative analysis are required

12. DATA ANALYSIS
Use of computers in data
analysis
Data storage and retrieval
Past and present

13. Importance of data analysis
Four major measurement scales
Nominal, Ordinal, Interval, Ratio
Designing tables
Graphic presentation
Data presentation
Frequency distribution
Central tendency
Normal curve and skewness
Standard deviation
Descriptive and inferential statistics
Parametric and non parametric tests
Chi – square, others

14. Qualitative data
Quantitative data
Differ in collection and processing

15. Preparation of data for analysis -
Review
Description of the sample
Testing the reliability of
measurements
Exploratory analysis
Confirmatory analysis
Post hoc analysis

16. Cleaning and organizing data for
computer entry
•Compilation-checking data for
accuracy, completeness and utility
•Editing –see that all questions are
answered and replies are internally
consistent

17. •Coding – transferring data into
symbols
•Classification of data
•Computer entry
•Making data backups

18. Obtain a complete picture of demographic
details/sample characteristics
Calculate measures of central tendency and
measures of dispersion
Comparison of various groups if more than
one group of sample

19. Reliability checked during the data collection
phase
need to be noted at this point.
If the reliability is unacceptably low –
( less than 0.7 for new tool and 0.8 for existing
tool )
you cannot justify performing analysis on the
data collected by these tools.

20. Examine data on each variable descriptively
Check whether the data is skewed or
normally distributed
Are there any outliers with extreme values?
Use of tables and graphs help to identify
patterns in the data and interpret
exploratory findings

21. It is performed to confirm researcher’s expectations
about the data that are expressed as research
questions, objectives or hypotheses.
Inferential statistical procedures
•Select level of significance used to interpret the results
•Choose the appropriate statistical test for analysis
•Determine the degrees of freedom
•Perform analysis manually or using computer
•Compare the obtained statistical value with the table value
•Interpret the results in terms of hypotheses, objectives or
questions

22. Usually performed in studies with more than
two groups
Helps to indicate how these groups are
related
eg. Analysis of variance
In other studies, insights obtained through
planned analysis can lead to further
questions that can be examined with the
available data

23. It is the search for broader meaning of the research
findings.
‘Translation’ is used synonym us with interpretation
where as
Translation means – transfer from one language to
other
Interpretation means
Explain meaning of information
Interpretation involves viewing findings
in light of available knowledge as Theory or
a Principle

24. An inference is the act of drawing conclusions
based on limited information, using logical
reasoning
The research findings are meant to reflect the
“Truth in the real world”
Study
results
Inference

Truth in the
real world

25. Interpreting study results involve attending to
six different but overlapping considerations
 The credibility and accuracy of results
 The precision of the estimate of effects
 The magnitude of effects and importance of
results
 The meaning of results, especially with regard
to causality
 The generalisability of results
 The implications of results for nursing practice,
theory development or further research

26. It requires inductive reasoning
Specific case to general truth
Pure – white
Concrete – abstract
Known and unknown
Increase probability that
Inferences are accurate
Inferences are made continuously and with
great care

27. Interpretations can be done by
Self scrutiny
Peer review
Outside review
Consider possible alternate explanations
take account of methodological and other
limitations that affect results
While drawing conclusions
Consider transferability of the findings
Do not generalize casually

28. Type
Statistical
Clinical
It requires
Abstract thinking
Introspection
Reasoning
Intuition
Interpretation has two aspects
Establish continuity in research
Establish explanatory concepts

29. Through
Unravel/understand the abstract principles beneath
the concrete observation
Link results with those of studies
Make prediction, raise questions
It is transit from exploratory to experimental
research
Findings of a study leads to hypothesis for the
future studies

30. Why observations/findings are /what they
are?
Thus theory assume central importance

31. Meaning of material
Implication of data
Conclusion
Recommendations
Values of data
Generalization
Comparison
Whitney, frederick lamon: the elements of
research ed 3; 410 Prentice – Hall inc 1950.

32. Logic used in plan
Examine findings
Form conclusions
Explore significance
Generalize findings
Consider implications
Suggest further studies

33. For interpretation of your research
Examine evidence
Critique your work
Synthesize judgment
It is time for not only
Confession
Remerge
Apology
but thoughtful reflection
Contd.

34. For interpretation of your research
--------------Contd.
Examine evidence from
Plans
Tools
Data collection
Data analysis
Data report
Significant results
Non – significant results
Mixed results
Unexpected results
Results of previous studies

35. Validity is in ‘degree’ / no absolute
Not proved or established
Supported by evidence
Validity is of application
and not of instrument

36. Interpretation of validity -----contd
Qualitative data quality
Evaluate trust worthiness of

Data

Interpretations
By using

Credibility of data

Dependability

Conformability

Transferability
Contd..

37. Credibility :refers to confidence in the truth of
data and its interpretation
Dependability :refers to the stability (reliability) of
data over time and over conditions
Conformability :refers to objectivity ie. Potential
for congruence between two or more independent
people about the data’s accuracy, relevance or
meaning
Transferability :refers to the extend to which the
findings are applicable to other settings or groups

38. Interpretation of validity -----contd
Credibility of data
Tools are
Peer briefings
Training and experience
Negative case analysis
Dependability
Over time
Over conditions
Conformability
Inquiry audits

39. Interpretation of validity -----contd
Credibility of data
Transferability
Through
Thick description
Context of data collection

40. Interpretations and analysis occur simultaneously
Researcher
Interpret data or categorize it
Develop thematic analysis
Integrate theme into unifying whole
Validate analysis
Validate interpretations

41. Qualitative data analysis involve a range of
procedures to get an explanation,
understanding or interpretation of people
and situations being studied.
It is intensive time consuming activity that
involve clustering together of qualitative
data.

42. There is no universal rule/method for
analyzing qualitative data
Enormous amount of work to draw
meaningful conclusions
Requires powerful inductive skills and
creativity
Qualitative data cannot be condensed too
much like quantitative data

43. Transcribing qualitative data
Developing category scheme
Coding data categories
Summarizing data in compilation sheets
Further summarizing of data in-
matrices/flow charts/diagrams/quasi-
statistical tables or narrative analysis
Drawing and verifying conclusions
Reporting results

44. Evaluate trust worthiness of

Data

Interpretations
By using

Credibility

Dependability

Confirmability

Transferability

45. Strategies related to data collection:
 Prolonged engagement and persistent
observation
 Data and method triangulation
 Comprehensive and vivid recording of
information
 Member checking

46. Strategies related to analysis:
 Investigator and theory triangulation
 Searching for disconfirming evidence
 Peer debriefing
 Inquiry audit
Strategies related to presentation:
Thick and contextualized description
Researcher credibility

47. In qualitative data analysis and
interpretation-
Consider possible alternate explanations on
account of methodological and other
limitations that affect results
While drawing conclusions:
Consider transferability of the findings
Do not generalize casually