The Student`s t-test

TheStudent’s t-test

  1. Description of the statistic you intend to examine in the article

TheStudent’s t-test

Thestudent’s t-statistic is a statistical method used to determinewhether two sets of data differ significantly. This method is basedon the assumption that if the null hypothesis is true, then theresults follow a normal distribution (t-distribution). The nullhypothesis states that there is no significant statistical differencebetween the means of two data sets. In statistics, t-statistics areused to determine the difference between two sample groups that areindependent. The t-value will be positive in cases where the firstmean is greater than the second mean and vice versa.

Afterobtaining the t-value, one has to look up the value in the table ofsignificance to determine the statistical difference between therespective samples. One has to set a risk level/level ofsignificance/alpha level n order to determine the significance. Inmany cases, the default alpha level is 0.05. Additionally, one mustdetermine the degrees of freedom (df) for the test. This is simplythe sum of observations in both samples minus two. With the alphalevel, the t-value and the degrees of freedom, it is possible to lookup the t-statistic in the standard table of significance. Thefollowing assumptions are made when conducting statistical tests:

  1. The data has a normal distribution

  2. The data sets have homogenous variances

  3. The data has a linear relationship

  4. The data is independent

  1. Description of the study portrayed in the article

Research ACROSS-CULTURAL STUDY OF THE CONCEPT OF CARING THROUGH BEHAVIORS:PATIENTS AND NURSES’ PERSPECTIVES IN SIX DIFFERENT EU COUNTRIES

Thepaper aims to investigate the global perceptions of patients andnurses on nurse caring behaviors. The authors appreciate the numerouscomparative studies exploring the perceptions of nurses and patientson nurses’ caring behaviour in different settings. However, thestudies have been contradictory, with the majority showingsignificant variations between the perceptions of patients and theperceptions of nurses. The studies reported that patients put morevalue on technical and instrumental caring skills than nurses do.Nurses, on the other hand, perceive their affective or expressivecaring behavior and psychological skills as more important to thepatients. Therefore, it is very likely that nurses may fail to assessaccurately patient perceptions of caring and thus not congruent topatient expectations.

Theresearch design

Thestudy used a descriptive, comparative study design. The researcherscollected data nurses and their patients admitted to surgical wardsin six countries (Cyprus, Greece, the Czech Republic, Italy, Finlandand Italy).

Theparticipants and Data Collection

Theresearchers collected data using participant-completed questionnairesgiven to inpatients and their nurses in the six countries. The studyemployed power analysis to determine the sample size using the NQueryAdvisor statistical software. The study required 150 nurses and 223patients to fill the questionnaires from each country to achieve a90% power level (α=0.01). The study had a response rate of 84.17%among patients and 76.26% among nurses. The questionnaires includedthe demographics and Caring Behaviour Inventory (CBI). Nurses andpatients can use the same version of the CBI, therefore, making iteasier to collect relevant data.

TheCBI forms were translated into the languages of the countriesparticipating in the research with equivalent semantic scales. Theresearcher appointed contact persons in each setting to distributethe questionnaires. The study required the participating partners tofollow their national guidelines on research settings, especiallyaccessing research settings.

Dataanalysis

SPSSversion 16.0 was used to analyze the data. The researchers analyzedbackground variables, scales, and items using descriptive statistics,frequencies, standard deviations, means and percentages. Theperceptions of nurses and patients about caring behaviors werecompared using independent samples t-test (p-value, t-statistics).The backgrounds of the nurses and patients were compared using theone-way analysis of variance.

  1. How the statistic was used in the study

Theresearchers analyzed the background variables, scales, and itemsusing descriptive statistics, standard deviations, means percentagesand frequencies. They made comparisons using inferential statistics.They compared perceptions of the nurses and patients of caringbehavior using independent samples of t-test (t-statistics, p-value).T-tests were used to show important differences in the first factor(Assurance of human presence) at a p-value &lt 0.001 and the thirdfactor (respectful deference to others) at p-value &lt 0.001. Thet-statics helped to show that the patients had lower means (feweranswers towards agreeing/strongly) compared to that of nurses.Independent samples t-tests showed the existence of importantdifferences between the nurses and the patients’ mean values forthe entire scale were only observed in the Czech Republic and Cyprus.

  1. Discussion of whether this is appropriate or not

Carryingout the t-tests was appropriate because, in this study, theresearchers were comparing two populations in every country coveredby the study i.e. the patients and the nurses. The data collectedcame from two independent groups, nurses, and patients. Additionally,comparing similar aspects in different countries, for example, theperceptions of nurses in Cyprus and the Czech Republic required theresearchers to carry out t-tests to identify any significantdifferences. Nurses and patients came from different geographicalareas, each representing an independent population.

  1. Explanation of how assumptions of the test were met or not met

Thefirst assumption is that the observations used in the t-test areindependent. This assumption was met because the observations werecollected from two independent group i.e. patients and nurses.Additionally, we have separate samples of observations from eachcountry under the study. The data collection was done independentlyin each country, therefore, the observations are 100% independent.

Thesecond assumption is that the observation/data has a normaldistribution. The researchers collected data from 150 nurses and 233patients in each country. These samples had more than 30observations. Therefore, in such a case, it is correct to assume thatthe data collected had a normal distribution. There was no need,therefore, to test for the normality of the data since this was alarge sample. This assumption was also met adequately.

Thethird assumption is that the standard deviations of the samples aresimilar. Since the researchers were trying to compare the means ofdifferent groups, it was paramount for the observations to havesimilar standard deviations for the t-test to work. This assumptionwas met in the study. All the three assumptions under t-test were metin the study. The study computed standard deviations on the nursesand patients’ profiles in all the countries. Therefore, we canconclude that the results of the study were conclusive.

  1. Identification of the levels of measurement of variables in the study

Thestudy used the interval level of measurement. The researchers linkedeach item in the CBI-24 to a 6-point Likert-type scale (1 = Never to6 = Always).

  1. Description of the appropriateness of the level of measurement

Thelevel of measurement was appropriate because the distance between theattributes have meaning. A higher mean response implied that a higherfrequency of mean is perceived. This would make it possible tocompute averages for the observations made in the study.

  1. Discussion of how the data was displayed

Thestudy displayed the data results in tables. Table1 contained a comparison of nurses and patients in the four factorsof CBI-24 (factor scores range from 1 to 6)

CBI-24

Group

n

Mean

SD

Dif

t-statist

P value

F1:Assurance of Human Presence

patients

1441

4.96

0.85

0.14

4.81

&lt0.001

Nurses

1099

5.10

0.68

F2:Knowledge and Skill

Patients

1448

5.30

0.78

0.01

0.51

0.608

Nurses

1111

5.29

0.63

F2:Knowledge and Skill

Patients

1413

4.72

0.98

0.15

4.11

&lt0.001

Nurses

1089

4.87

0.77

F4:Positive Connectedness

Patients

1472

4.63

1.02

0.05

1.32

0.188

Nurses

1108

4.58

0.80

Differenceis statistically significant at the 0.01 level

Table2 contained the estimated marginal means*, confidence intervals,ANCOVA results (F-Statistic, degrees of freedom, P-value), forcross-country comparisons for the CBI-24 scale used in obtaining thedata.

Nurses

Mean (95% CI)

F (d.f)

P value

Patients

Mean (95% CI)

F (d.f.)

P value

Cyprus

4,69(4.57, 4.82)

24.199 (5)

&lt0.001

Cyprus

5.03 (4.90, 5.17)

26.945 (5)

&lt0.001

Italy

5.04 (4.94, 5.14)

Italy

4.87 (4.73, 5.01)

Hungary

5.23 (5.12, 5.33)

Hungary

5.30 (5.18, 5.43)

Czech Republic

5.06 (4.96, 5.16)

Czech Republic

4.67 (4.56, 4.78)

Greece

4.52 (4.42, 4.26)

Greece

4.48 (4.36, 5.58)

Finland

5.08 (5.00, 5.16)

Finland

5.13 (5.02, 5.24)

Estimatedthrough the general linear model, ANCOVA

Table3 showed a summary of the Nurse–Patient differences per country.

  1. Discussion examining the appropriateness of the data displays

Presentingthe data is tables was appropriate because it gives readers easyvisualization of the summarized data and results of the study. Thedescription of the data was more poignant with the adoption of thetables.